&EPA
United States
Environmental Protection
Agency
Office of Research and
Development
Washington DC 20460
EPA/600/8-90/057B6
December 1994
External Review Draft
Health Assessment
Document for
Diesel Emissions
Volume II of II
Review
Draft
(Do Not
Cite or
Quote)
Notice
This document is a preliminary draft. It has not been formally
released by EPA and should not at this stage be construed to
represent Agency policy. It is being circulated for comment on its
technical accuracy and policy implications.
-------
DRAFT-DO NOT QUOTE OR CITE
EPA/600/8-90/057Bb
December 1994
External Review Draft
Health Assessment Document
for Diesel Emissions
Volume II of II
NOTICE
This document Is a preliminary draft. It has not been formally
released by EPA and should not at this stage be construed to
represent Agency policy. It Is being circulated for comment on
its technical accuracy and policy implications.
Environmental Criteria and Assessment Office
Office of Health and Environmental Assessment
Office of Research and Development
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
U.S. Environmental Protection Agency
Region 5, Library (PL-12J) <6$ Printed on Recycled Paper
77 West Jackson Boulevard, 12th Floor
Chicago, IL 60604-3590
-------
DISCLAIMER
This document is an external draft for review purposes only and does not constitute
U.S. Environmental Protection Agency policy. Mention of trade names or commercial
products does not constitute endorsement or recommendation for use.
December 1994 H-ii DRAFT-DO NOT QUOTE OR CITE
-------
Health Assessment Document for Diesel Emissions
TABLE OF CONTENTS
Volume I Page
1. EXECUTIVE SUMMARY 1-1
2. DIESEL EMISSIONS 2-1
3. DIESEL-DERIVED POLLUTANTS: ATMOSPHERIC
CONCENTRATIONS, TRANSPORT, AND TRANSFORMATIONS . . 3-1
4. DOSIMETRIC FACTORS 4-1
\T\ 5. NONCANGER HEALTH EFFECTS OF DIESEL EXHAUST 5-1
r\
\S> 6. QUALITATIVE AND QUANTITATIVE ASSESSMENT OF
tV\ NONCANCER HEALTH-EFFECTS-DERIVATION OF THE
INHALATION REFERENCE CONCENTRATION 6-1
V)
\) Volume II
^
, 7. CARCINOGENICITY OF DIESEL EMISSIONS IN LABORATORY
^P ANIMALS 7-1
^
8. EPIDEMIOLOGIC STUDIES OF THE CARCINOGENICITY OF
EXPOSURE TO DIESEL EMISSIONS 8-1
Addendum to Chapter 8 8A-1
9. MUTAGENICITY 9-1
10. METABOLISM AND MECHANISM OF ACTION IN DIESEL
EMISSION-INDUCED CARCINOGENESIS 10-1
11. QUALITATIVE AND QUANTITATIVE EVALUATION OF THE
CARCINOGENICITY OF DIESEL ENGINE EMISSIONS 11-1
12. HEALTH RISK CHARACTERIZATION FOR DIESEL ENGINE
EMISSIONS 12-1
APPENDIX A: EXPERIMENTAL PROTOCOL AND COMPOSITION OF
EXPOSURE ATMOSPHERES A-l
December 1994 H-iii DRAFT-DO NOT QUOTE OR CITE
-------
TABLE OF CONTENTS (cont'd)
APPENDIX B: ASSESSMENT OF RISK FROM EXPOSURE TO DIESEL
ENGINE EMISSIONS B-l
APPENDIX C: ALTERNATIVE MODEL FOR DIESEL CANCER RISK
ASSESSMENT C-l
APPENDIX D: MODELS FOR CALCULATING LUNG BURDENS .... D-l
December 1994 H-iv DRAFT-DO NOT QUOTE OR CITE
-------
TABLE OF CONTENTS
VOLUME II
Pas
LIST OF TABLES II-x
LIST OF FIGURES Il-xii
AUTHORS, REVIEWERS, AND CONTRIBUTORS Il-xiii
ACKNOWLEDGMENTS Il-xvi
7. CARCINOGENICITY OF DIESEL EMISSIONS IN LABORATORY
ANIMALS 7-1
7.1 INTRODUCTION 7-1
7.2 INHALATION STUDIES 7-2
7.2.1 Rat Studies 7-2
7.2.2 Mouse Studies 7-18
7.2.3 Hamster Studies 7-22
7.2.4 Monkey Studies 7-25
7.3 LUNG IMPLANTATION OR INTRATRACHEAL
INSTILLATION STUDIES 7-25
7.3.1 Rat Studies 7-25
7.3.2 Syrian Hamster Studies 7-28
7.4 SUBCUTANEOUS AND INTRAPERITONEAL INJECTION
STUDIES 7-29
7.4.1 Mouse Studies 7-29
7.5 DERMAL STUDIES 7-31
7.5.1 Mouse Studies 7-31
7.6 SUMMARY AND CONCLUSIONS OF ANIMAL
CARCINOGENICITY STUDIES 7-37
REFERENCES 7-45
8. EPIDEMIOLOGIC STUDIES OF THE CARCINOGENICITY OF
EXPOSURE TO DIESEL EMISSIONS 8-1
8.1 INTRODUCTION 8-1
8.2 COHORT STUDIES 8-2
8.2.1 Waller (1981): Trends in Lung Cancer in London
in Relation to Exposure to Diesel Fumes 8-2
8.2.2 Howe et al. (1983): Cancer Mortality (1965 to 1977)
in Relation to Diesel Fume and Coal Exposure in a
Cohort of Retired Railroad Workers 8-4
8.2.3 Rushton et al. (1983): Epidemiological Survey of
Maintenance Workers in the London Transport Executive
Bus Garages and Chiswick Works 8-6
8.2.4 Wong et al. (1985): Mortality Among Members of a
Heavy Construction Operators Union with Potential
Exposure to Diesel Exhaust Emissions 8-7
December 1994 n.v DRAFT-DO NOT QUOTE OR CITE
-------
TABLE OF CONTENTS (cont'd)
8.2.5 Edling et al. (1987): Mortality Among Personnel
Exposed to Diesel Exhaust 8-12
8.2.6 Boffetta and Stellman (1988): Diesel Exhaust Exposure
and Mortality Among Males in the American Cancer
Society Prospective Study 8-13
8.2.7 Garshick et al. (1988): A Retrospective Cohort Study of
Lung Cancer and Diesel Exhaust Exposure in Railroad
Workers 8-15
8.3 CASE-CONTROL STUDIES OF LUNG CANCER 8-19
8.3.1 Williams et al. (1977): Associations of Cancer Site
and Type with Occupation and Industry from the Third
National Cancer Survey Interview 8-19
8.3.2 Hall and Wynder (1984): A Case-Control Study of
Diesel Exhaust Exposure and Lung Cancer 8-24
8.3.3 Damber and Larsson (1987): Occupation and Male
Lung Cancer, a Case-Control Study in Northern
Sweden 8-25
8.3.4 Lerchen et al. (1987): Lung Cancer and Occupation
in New Mexico 8-27
8.3.5 Garshick et al. (1987): A Case-Control Study of
Lung Cancer and Diesel Exhaust Exposure in Railroad
Workers 8-29
8.3.6 Benhamou et al. (1988): Occupational Risk Factors
of Lung Cancer in a French Case-Control Study 8-33
8.3.7 Hayes et al. (1989): Lung Cancer in Motor
Exhaust-Related Occupations 8-34
8.3.8 Steenland et al. (1990): A Case-Control Study of Lung
Cancer and Truck Driving in the Teamsters Union .... 8-36
8.4 CASE-CONTROL STUDIES OF BLADDER CANCER 8-38
8.4.1 Howe et al. (1980): Tobacco Use, Occupation,
Coffee, Various Nutrients, and Bladder Cancer 8-38
8.4.2 Wynder et al. (1985): A Case-Control Study of Diesel
Exhaust Exposure and Bladder Cancer 8-44
8.4.3 Hoar and Hoover (1985): Truck Driving and Bladder
Cancer Mortality in Rural New England 8-46
8.4.4 Steenland et al. (1987): A Case-Control Study of
Bladder Cancer Using City Directories as a
Source of Occupational Data 8-48
8.4.5 Iscovich et al. (1987): Tobacco Smoking, Occupational
Exposure, and Bladder Cancer in Argentina 8-50
8.4.6 Iyer et al. (1990): Diesel Exhaust Exposure and
Bladder Cancer Risk 8-53
8.4.7 Steineck et al. (1990): Increased Risk of Urothelial
Cancer in Stockholm from 1985 to 1987, after Exposure
to Benzene and Exhausts 8-54
December 1994 n-vi DRAFT-DO NOT QUOTE OR CITE
-------
TABLE OF CONTENTS (cont'd)
8.5 DISCUSSION AND SUMMARY 8-56
8.5.1 The Cohort Mortality Studies 8-62
8.5.2 Case-Control Studies of Lung Cancer 8-65
8.5.3 Case-Control Studies of Bladder Cancer 8-67
8.5.4 Relevant Methodologic Issues 8-67
8.5.5 Criteria of Causal Inference 8-68
REFERENCES 8-72
Addendum to Chapter 8 8A-1
9. MUTAGENICITY 9-1
9.1 GENE MUTATIONS 9-1
9.2 CHROMOSOME EFFECTS 9-4
9.3 OTHER GENOTOXIC EFFECTS 9-6
9.4 SUMMARY 9-6
REFERENCES 9-9
10. METABOLISM AND MECHANISM OF ACTION IN DIESEL
EMISSION-INDUCED CARCINOGENESIS 10-1
10.1 METABOLISM AND MECHANISM OF ACTION OF
ORGANIC CARCINOGENIC COMPONENTS OF DIESEL
EXHAUST 10-1
10.1.1 Metabolism and Disposition of Benzo[fl]pyrene 10-2
10.1.2 Carcinogenic Mechanism of Benzo[a]pyrene 10-7
10.1.3 Metabolism and Disposition of Nitroarenes 10-9
10.1.4 Carcinogenic Mechanism of Nitroarenes 10-15
10.2 PARTICLE EFFECT IN DIESEL EXHAUST-INDUCED
CARCINOGENICITY 10-20
10.3 POTENTIAL INVOLVEMENT OF PULMONARY
LEUKOCYTES IN THE DEVELOPMENT OF LUNG
TUMORS 10-22
10.4 MOLECULAR DOSIMETRY CONSIDERATIONS 10-27
10.5 SUMMARY OF METABOLISM AND MECHANISM OF
ACTION OF CARCINOGENIC COMPONENTS OF
DIESEL EXHAUST 10-29
REFERENCES 10-33
11. QUALITATIVE AND QUANTITATIVE EVALUATION OF THE
CARCINOGENICITY OF DIESEL ENGINE EMISSIONS 11-1
11.1 INTRODUCTION 11-1
11.2 WEIGHT OF EVIDENCE FOR CARCINOGENICITY OF
DIESEL EXHAUST 11-2
11.3 REVIEW OF PREVIOUS QUANTITATIVE RISK
ESTIMATES 11-4
11.4 APPROACHES TO QUANTITATION OF HUMAN RISK
FROM EXPOSURE TO DIESEL EXHAUST 11-10
December 1994 H-vii DRAFT-DO NOT QUOTE OR CITE
-------
TABLE OF CONTENTS (cont'd)
11.5 DOSE-RESPONSE CALCULATIONS BASED ON ANIMAL
BIOASSAY DATA 11-15
11.5.1 Data Available for Risk Calculations 11-15
11.5.2 Calculation of Unit Risks 11-17
11.5.3 Results of Unit Risk Calculations 11-22
11.5.4 Discussion of Unit Risk Estimates 11-23
11.5.4.1 Basis for the Present Approach 11-23
11.5.4.2 Evaluation of Animal-Based Risk Estimates
Against Human Experience 11-33
11.5.4.3 Reasonableness of the Unit Risk Estimate . . 11-36
11.6 SUMMARY AND CONCLUSIONS 11-38
REFERENCES 11-39
12. HEALTH RISK CHARACTERIZATION FOR DIESEL ENGINE
EMISSIONS 12-1
12.1 INTRODUCTION 12-1
12.2 ACUTE EXPOSURE HAZARDS 12-2
12.2.1 Hazard Identification 12-2
12.2.2 Dose Response for Acute Toxicity 12-2
12.3 CHRONIC NONCARCINOGENIC EXPOSURE HAZARDS . . . 12-2
12.3.1 Hazard Identification 12-2
12.3.2 Dose Response for Chronic Toxicity 12-3
12.3.2.1 Selection of Dose-Response Data 12-3
12.3.2.2 Dose Measure 12-3
12.3.2.3 Dose Equivalence Across Species 12-4
12.3.2.4 Inhalation Reference Concentration
Derivation 12-4
12.3.2.5 Reasonableness and Utility of the
Inhalation Reference Concentration 12-4
12.4 CARCINOGENIC EXPOSURE HAZARDS 12-5
12.4.1 Hazard Identification 12-5
12.4.2 Methods for Determining Dose Response 12-6
12.4.2.1 Selection of Dose-Response Data 12-6
12.4.2.2 Dose Measure 12-7
12.4.2.3 Dose Equivalence Across Species 12-8
12.4.2.4 High-to-Low-Dose Risk Extrapolation 12-9
12.4.3 Results of Dose-Response Calculations 12-11
12.4.3.1 Results Using the Linearized Multistage
Model 12-11
12.4.3.2 Results Using the Alternative Model 12-12
12.4.4 Discussion of Confidence in the Upper Bound
Risk Estimates 12-12
12.5 EXPOSURE ESTIMATES 12-15
12.5.1 Methodology 12-15
12.5.2 Confidence in Exposure Estimates 12-16
December 1994 H-viii DRAFT-DO NOT QUOTE OR CITE
-------
TABLE OF CONTENTS (cont'd)
12.6 POPULATION RISKS AND UNCERTAINTIES 12-18
12.6.1 Population Risks for the Induction of Noncancer
Toxicity 12-18
12.6.1.1 Population Risks for Acute Exposure 12-18
12.6.1.2 Population Risks for Chronic Exposure .... 12-18
12.6.2 Population Risks for Induction of Cancer 12-20
12.6.3 Comparison of Cancer and Noncancer Risk Estimates .. 12-21
12.7 SUMMARY 12-22
REFERENCES 12-24
APPENDIX A: EXPERIMENTAL PROTOCOL AND COMPOSITION
OF EXPOSURE ATMOSPHERES A-l
APPENDIX B: ASSESSMENT OF RISK FROM EXPOSURE TO
DIESEL ENGINE EMISSIONS B-l
APPENDIX C: ALTERNATIVE MODEL FOR DIESEL CANCER
RISK ASSESSMENT C-l
APPENDIX D. MODELS FOR CALCULATING LUNG BURDENS . . D-l
December 1994 H_ix DRAFT-DO NOT QUOTE OR CITE
-------
LIST OF TABLES
Number Page
7-1 Summary of Animal Carcinogenicity Studies 7-3
7-2 Tumor Incidence and Survival Time of Rats Treated with
Fractions from Diesel Exhaust Condensate 7-26
7-3 Tumorigenic Effects of Dermal Application of Acetone
Extracts of Diesel Exhaust 7-33
7-4 Dermal Tumorigenic and Carcinogenic Effects of Various
Emission Extracts 7-36
7-5 Cumulative Exposure Data for Rats Exposed to Whole
Diesel Exhaust 7-40
8-1 Epidemiologic Studies of Health Effects of Exposure to
Diesel Exhaust: Cohort Mortality Studies 8-20
8-2 Epidemiologic Studies of Health Effects of Exposure to
Diesel Exhaust: Case-Control Studies of Lung Cancer 8-39
8-3 Epidemiologic Studies of Health Effects of Exposure to
Diesel Exhaust: Case-Control Studies of Bladder Cancer 8-57
11-1 Estimated Lifetime Risk of Cancer from Inhalation
of 1 /xg/m3 Diesel Paniculate Matter 11-9
11-2 Incidence of Lung Tumors in Fischer 344 Rats Exposed
to Heavy-Duty Engine Exhaust 11-16
11-3 Incidence of Lung Tumors in Fischer 344 Rats Exposed
to Heavy-Duty Engine Exhaust 11-16
11-4 Incidence of Lung Tumors in Fischer 344 Rats Exposed
to Diesel Engine Exhaust 11-17
11-5 Unit Risk Estimates per Microgram per Cubic Meter of
Diesel Exhaust 11-22
11-6 Cancer Studies with Rats Exposed to Relatively
Chemically Inert Dusts at Exposure Concentrations
of Several Micrograms per Cubic Meter or Above 11-33
December 1994 H_x DRAFT-DO NOT QUOTE OR CITE
-------
LIST OF TABLES (cont'd)
Number
12-1 Unit Risk Estimates per Micrograms per Cubic Meter of Diesel
Exhaust 12-11
12-2 Estimated Annual Ambient Concentrations of Diesel
Exhaust Paniculate Matter 12-18
December 1994 n_xi DRAFT-DO NOT QUOTE OR CITE
-------
LIST OF FIGURES
Number Page
7-1 Cumulative exposure data for rats exposed to whole diesel
exhaust 7-42
11-1 Calculated lung burden (organic matter) in rats exposed
to three different concentrations of paniculate matter 11-19
11-2 Calculated lung burden (carbon core) in rats exposed
to three different concentrations of paniculate matter 11-20
11-3 Relationship between lung tumor incidence and modeled
lung particle burden/unit of lung surface area using data
from Brightwell et al. (1986), Ishinishi et al. (1986),
and Mauderly et al. (1987) 11-30
11-4 Relationship between exposure rate and lung particle
burden/unit of lung surface area using data from
Brightwell et al. (1986), Ishinishi et al. (1986), and
Mauderly et al. (1987) 11-31
12-1 Relationship of exposure estimates and risk-specific doses 12-19
December 1994 H-xii DRAFT-DO NOT QUOTE OR CITE
-------
AUTHORS, REVIEWERS, AND CONTRIBUTORS
Authors
Dr. Ronald Bradow
Atmospheric Research and Exposure
Assessment Laboratory
U.S. Environmental Protection Agency
Research Triangle Park, NC
Dr. Chao Chen
Human Health Assessment Group
U.S. Environmental Protection Agency
Washington, DC
Dr. Kenney Crump
Clement International Corporation
Ruston, LA
Dr. Daniel Guth
Environmental Criteria and Assessment
Office
U.S. Environmental Protection Agency
Research Triangle Park, NC
Dr. John Johnson
Houghton, MI
Dr. Aparna Koppikar
Human Health Assessment Group
U.S. Environmental Protection Agency
Washington, DC
Ms. Tammie Lambert
Clement International Corporation
Ruston, LA
Dr. Bruce Lehnert
Pulmonary Biology-Toxicology Section
Los Alamos National Laboratory
Los Alamos, NM
Dr. Samuel Lestz
State College, PA
Dr. Kumar Menon
Pikesville, MD
Dr. Giinter Oberdorster
Department of Biophysics
University of Rochester Medical Center
Rochester, NY
Dr. Dennis Opresko
Biomedical and Environmental Information
Analysis
Health and Safety Research Division
Oak Ridge National Laboratory
Oak Ridge, TN
Dr. William Pepelko
Human Health Assessment Group
U.S. Environmental Protection Agency
Washington, DC
Mr. Chris Rambin
Clement International Corporation
Ruston, LA
Dr. Larry Valcovic
U.S. Environmental Protection Agency
Washington, DC
Mr. Michael Walsh
Arlington, VA
Dr. Ronald K. Wolff
Lilly Research Laboratories
Greenfield, IN
Dr. Robert A. Young
Biomedical and Environmental Information
Analysis
Health Sciences Research Division
Oak Ridge National Laboratory
Oak Ridge, TN
December 1994
DRAFT-DO NOT QUOTE OR CITE
-------
AUTHORS, REVIEWERS, AND CONTRIBUTORS (cont'd)
Authors (cont'd)
Dr. Chia-Ping Yu
State University of New York at Buffalo
Department of Mechanical and Aerospace
Engineering
Buffalo, NY
Dr. Barbara Zielinska
Desert Research Institute
Energy and Environmental Engineering
Center
Reno, NV
Project Managers
Mr. William Ewald
Environmental Criteria and Assessment
Office
U.S. Environmental Protection Agency
Research Triangle Park, NC
Dr. William Pepelko
Office of Health and Environmental
Assessment
U.S. Environmental Protection Agency
Washington, DC
Reviewers and Contributors
The following individuals reviewed the current and/or an earlier draft of this document and
participated in a peer review workshop on July 18 and 19, 1990.
Dr. Roy Albert
University of Cincinnati
Cincinnati, OH
Dr. James Bond
Chemical Industries Institute of
Toxicology
Research Triangle Park, NC
Dr. Glen Cass
California Institute of Technology
Pasadena, CA
Dr. Eric Garshik
Harvard Medical School
Channing Laboratory
Boston, MA
Dr. Judith Graham
Environmental Criteria and Assessment
Office
U.S. Environmental Protection Agency
Research Triangle Park, NC
Dr. Uwe Heinrich
Department of Environmental Hygiene
Fraunhofer Institute
Hanover, Germany
Dr. Joellen Lewtas
Health Effects Research Laboratory
Research Triangle Park, NC
Dr. Joe Mauderly
Lovelace Inhalation Research Institute
Albuquerque, NM
December 1994
Il-xiv
DRAFT-DO NOT QUOTE OR CITE
-------
AUTHORS, REVIEWERS, AND CONTRIBUTORS (cont'd)
Reviewers and Contributors (cont'd)
Dr. Roger McClellan
Chemical Industries Institute of
Toxicology
Research Triangle Park, NC
Dr. Fred Miller
Duke University Medical Center
Durham, NC
Dr. Otto Raabe
University of California
Davis, CA
Mr. Charles Ris
Human Health Assessment Group
U.S. Environmental Protection Agency
Washington, DC
Dr. Herbert Rosenkranz
Department of Environmental Sciences
School of Medicine
Case Western University
Cleveland, OH
Dr. Irving Salmeen
Ford Motor Company
Scientific Research Lab
Dearborn, MI
Dr. Andrew Sivak
Cambridge, MA
Dr. Jeanette Wiltse
Office of Health and Environmental
Assessment
U.S. Environmental Protection Agency
Washington, DC
Dr. Thomas Smith
University of Massachusetts Medical
Center
Worchester, MA
Dr. Frank Speizer
Charming Laboratory
Boston, MA
Dr. Leslie Stayner
National Institute for Occupational Safety
and Health - Taft Labs
Cincinnati, OH
Dr. Werner Stober
Chemical Industries Institute of
Toxicology
Research Triangle Park, NC
Dr. Jaroslav Vostal
General Motors Research Labs
Warren, MI
December 1994
II-xv
DRAFT-DO NOT QUOTE OR CITE
-------
ACKNOWLEDGMENTS
Word Processing Support
Ms. Glenda Johnson
Biomedical and Environmental
Information Analysis
Health Sciences Research Division
Oak Ridge National Laboratory
Oak Ridge, TN
Document Production
Ms. Marianne Barrier
Mr. John Barton
Ms. Sheila Lassiter
Ms. Wendy Lloyd
Ms. Edie Smith
ManTech Environmental Technology, Inc.
Research Triangle Park, NC
References
Mr. Douglas Fennell
Environmental Criteria and Assessment
Office
U.S. Environmental Protection Agency
Research Triangle Park, NC
Ms. Catherine Carter
Ms. Blythe Hatcher
Information Organizers, Inc.
Research Triangle Park, NC
Reprographics
Mr. Richard Wilson
Environmental Criteria and Assessment
Office
U.S. Environmental Protection Agency
Research Triangle Park, NC
December 1994
H-xvi DRAFT-DO NOT QUOTE OR CITE
-------
i 7. CARCINOGENICITY OF DIESEL EMISSIONS IN
2 LABORATORY ANIMALS
3
4
5 7.1 INTRODUCTION
6 The paniculate phase of diesel exhaust is composed of aggregates of carbon particles;
7 the primary particle diameter ranges from 10 to 80 nm, and aggregates of these primary
8 particles have mass median diameters averaging 0.2 to 0.3 /urn (Vuk et al., 1976; Carpenter
9 and Johnson, 1980), although some may approach 1.0 j«m. A great variety of organic
10 compounds, including polycyclic aromatic hydrocarbons (PAHs), are adsorbed to this carbon
11 core (see Tables 2-6 and 2-8) and comprise 5 to 65 % of the total particle mass (Cuddihy
12 et al., 1984). Some of these organic compounds, such as benzo[a]pyrene (B[a]P),
13 dinitropyrenes, and 1-nitropyrene, have received special attention regarding their
14 carcinogenic and mutagenic potential. These organics may be strongly or weakly bound to
15 the carbon core and represent varying amounts of the total particle mass. Qualitative and
16 quantitative relationships for these organics depend on such variables as fuel composition,
17 engine design, and engine operating conditions. Although less emphasis has been placed on
18 the gaseous phase, potential carcinogens such as formaldehyde, acetaldehyde, and benzene,
19 as well as lower molecular weight PAHs, may be also be present in this fraction.
20 The respirability of these particles and their associated organics provides a basis for
21 health hazard concerns, and the reported mutagenicity (Huisingh et al., 1978) and skin
22 papilloma induction (Kotin et al., 1955) of solvent extracts of diesel soot suggests a potential
23 for carcinogenicity. Zamora et al. (1983) provided evidence that diesel exhaust particle
24 extracts contained components that acted as weak tumor promoters in vitro. Recently,
25 emphasis has been directed toward assessing the carcinogenic potential of whole and filtered
26 diesel exhaust using whole-animal studies and understanding the mechanisms and implications
27 of deposition, retention, and clearance of the paniculate phase of diesel exhaust.
28 This chapter summarizes studies that assess the carcinogenic potential of diesel exhaust
29 in laboratory animals. Experimental protocols for the inhalation studies usually consisted of
30 exposure (usually chronic) to diluted exhaust in whole-body exposure chambers using rats,
31 mice, and hamsters as model species. Some of these studies used both filtered (free of
December 1994 7_1 DRAFT-DO NOT QUOTE OR CITE
-------
1 paniculate matter) diesel exhaust and unfiltered (whole) diesel exhaust to differentiate
2 gaseous-phase effects from effects induced by the particulate matter and its adsorbed
3 components. Inhalation exposure to particulate matter alone, however was not reported.
4 Particulate matter concentrations in the diesel exhaust used in these studies ranged from
5 0.1 to 12 mg/m3. Clean air (usually filtered) was used in the control exposures. Studies
6 providing both positive, negative, or inconclusive findings have been reported. In this
7 chapter, any indication of statistical significance implies that p < 0.05 was reported in the
8 reviewed publications. The experimental protocols and exposure atmosphere
9 characterizations are not described in detail here but may be found in Appendix A.
10 A summary of the animal carcinogenicity studies and their results are presented in Table 7-1.
11 Also included are studies that assessed the carcinogenic and/or tumorigenic effects of
12 diesel exhaust particles and solvent extracts of these particles following dermal application,
13 sc injection, ip injection, or intratracheal instillation in rodents, as well as cocarcinogenicity
14 studies. Individual chemicals present in the gaseous phase or adsorbed to the particle surface
15 were not included in this review because adequate assessments of those of likely concern
16 (i.e., formaldehyde, acetaldehyde, benzene, and PAHs) have been published in other health
17 assessment documents.
18
19
20 7.2 INHALATION STUDIES
21 7.2.1 Rat Studies
22 Mauderly et al. (1987) provided data affirming the carcinogenicity of automotive diesel
23 engine exhaust in F344/Crl rats following chronic inhalation exposure. Male and female rats
24 were exposed to diesel engine exhaust at nominal particulate matter concentrations of
25 0.35 (n = 366), 3.5 (n = 367), or 7.1 (n = 364) mg/m3 for 7 h/day, 5 days/week for up to
26 30 mo. Sham-exposed (n = 365) controls breathed filtered room air. A total of 230, 223,
27 221, and 227 of these rats (sham-exposed, low-, medium-, and high-exposure groups,
28 respectively) were examined for lung tumors. These numbers included those animals that
29 died or were euthanized during exposure and those that were terminated following 30 mo of
30 exposure. The exhaust was generated by 1980 Model 5.7-L Oldsmobile V-8 engines
31 operated through continuously repeating U.S. Federal Test Procedure (FTP) urban
December 1994 7_2 DRAFT-DO NOT QUOTE OR CITE
-------
1
g
03
H
C
HH
§
O
o
ARCIN
U
1
<|
ta
O
>*
%
Q
A
(/)
.
^H
ri
3
M
3
H
1
o
U
«
Incidence (%)
8,
o
B a
Postexpos
Observati
§•3
II
w £
s 1
o 2
w IM
c
ill
o ^~"
O
a 1.
If
.§
"a 3
*a
0
E-l
u
•"I
(§* ^
X
1
CO
u
OO '— - ° °
ei__u^ g £ g| 3. £ P! 222 %%
^jlcSa ^1|®» ^Jssjs ^s^ 5|
PI ^ -» H«S!g H -* 3; ^ ooo 3H
S S u; - «s
^ 1 (g-t=oSSv5'f:: c?t3
•HiffffRR i|§S2 |e|| §2§
ll'lBC-S-ti '1 S S o\ SSgJtS o^S^
G §- S rtlooo cr$oo?^ Boo
^ + ^c
Wl ^ «^^. W ^^^ -^^ >***•
| ~~"" |°°l |f^S Soo
< < < <
z z z z
j. ^ MH
4) 4) 4> 1>
4> 4i 4J * U
Ilii f!ii
sgss sss sss sss
oooo ooo ooo ooo
zzzz zzz zzz zzz
en v\ ^ q o q
oornt~ oo^t OO1*' oot
^j^..^,. ^*- ^*-» ^*j
tflMCA 3(A 3(A 3M
•3 u « « '« -a u '5 -a " '3 -a "
— i> t> a> _, S o Co Sw
« O O O c3uO TO w O « o O
•0 m - t-
mtsrsts Tfrcnvovo'ovo
rsfststs ooost^c>^o\
QU tLi UH tb fTi Tri fT-i t11! flj [Tt
*t~"T'"t~~t" ON SN o\ 4~ + + "4""H +
2 S 2 2 n." u. u," S S S 2 S S
| -g
1 f =2 |l
3 1 I z ffi w
1 5 • a
*o f^ 'S ^ *-" ^ 'S **S
3 oo .g oo ^ oo .g oo
ISC' ffid'SC' tec
OO OO *O OS OO OO
^ oo ^- 1^ en so
^
^
•«-" a)
(Q ^ (Q ^
e ^1 £ c -Si S
nj O o> rt O 4j
D s* £ 0 ? S
WJ t/3 W3 CO C/5 CO
Z Z Z Z Z Z
U. (b UH UH It, UL,
H
•^ .2
<2p?
73
u
J3 ^-.
•c <>
H oo
U ON
S3 C
December 1994
7-3
DRAFT-DO NOT QUOTE OR CITE
-------
O
8
I
n>
TABLE 7-1 (cont'd). SUMMARY OF ANIMAL CARCINOGENICITY STUDIES
\*/
$
O
O
25
O
H
O
d
o
H
W
g
G
H
W
Species/
Study Strain
Takaki et al. Rat/F344
(1988)
Light-duty
engine
Ishinishi et al. Rat/F344
(1988b)
Heavy-duty
engine
Iwai et al. Rat/F344
(1986)
Takemoto et Rat/F344
al. (1986)
Particle
Sex/ Exposure Concentration Other Exposure Postexposure
Total Number Atmosphere (mg/m3) Treatment Protocol Observation
M + F, 123
M + F, 123
M + F, 125
M + F, 123
M + F, 124
M + F, 123
M + F, 123
M + F, 125
M + F, 123
M + F, 124
F, 24
F, 24
F, 24
F, 12
F, 21
F, 15
F, 18
Clean air
Whole exhaust
Whole exhaust
Whole exhaust
Whole exhaust
Clean air
Whole exhaust
Whole exhaust
Whole exhaust
Whole exhaust
Clean air
Filtered exhaust
Whole exhaust
Clean air
Clean air
Whole exhaust
Whole exhaust
0
0.1
0.4
1.1
2.3
0
0.5
1.0
1.8
3.7
0
0
4.9
0
0
4.9
0
0
2-4
2-4
None
None
None
None
None
None
None
None
None
None
None
None
None
None
DIPNh
None
DIPNh
16 h/day, NA
6 days/week,
for up to
30 mo
16 h/day, NA
6 days/week,
for up to
30 mo
8 h/day, NA
7 days/week,
for 24 mo
4 h/day, NA
4 days/week,
18-24 mo
Adenomas
1/23 (0.8)
1/23 (0.8)
1/25 (0.8)
0/23 (0)
1/24(8.1)
Adenomas
0/123 (0)
0/123 (0)
0/125 (0)
0/123 (0)
0/124 (0)
Adenomas
1/22 (4.5)
0/16 (0)
3/19 (0)
Tumor Type and Incidence (%f
Adenosquamous
Carcinomas
2/123 (1.6)
1/23 (0.8)
0/125 (0)
5/123(4.1)
2/124(1.6)
Adenosquamous
Carcinomas
1/123 (0.8)
0/123 (0)
0/125 (0)
4/123 (3.3)
6/124 (4.8)
Adenocarcinoma and
Adeno-Squamous
Carcinoma
0/22 (0)
0/16 (0)
3/19(15.8)
Adenoma
0/12 (0)
10/21 (47.6)
0/15 (0)
12/18 (66.7)
Squamous Cell
Carcinomas
1/23 (0.8)
1/23 (0.8)
0/125 (0)
0/123 (0)
0/124 (0)
Squamous Cell
Carcinomas
0/123 (0)
1/123 (0.8)
0/125 (0)
0/123 (0)
2/124(1.6)
Large Cell and
Squamous Cell
Carcinomas
0/22 (0)
0/16 (0)
2/19 (10.5)
Carcinoma
0/12 (0)
4/21 (19)
0/15 (0)
7/18 (38.9)
Comments
All
Tumors
4/123 (3.3)
3/123 (2.4)
1/125 (0.8)
5/123(4.1)
3/124 (2.4)
All
Tumors
1/123 (0.8)
1/123 (0.8)
0/125 (0)
4/123 (3.3)
8/124
(6.5)c
All
Tumors
1/22 (4.5)f
0/16 (0)
8/19
-------
Tf\
1
pH
Cfl
C
F^
U
6
8
^
U
U
1
fc
^J
^
fa
o
5
s
K
NjJ
C^J
I
i-H
I
i
V™5
"^
P^N
1
£
c
o
U
«
teS
U
S
c
a
^
o
1
£ e
3 0
•a
o «
t|
1 o
Is
W
)-. 0
u K
o S
H
B
O
•5 !"£
5 g "Sb
<2 S S,
0
u
Exposure
Atmosphere
1
~x 3
u Z
t/5
a
o
'3 '«
1
4> (A
y "S /—v '"**
«3 C o w? ws
*rt -H 'C \rt ^
0 U §. ^ 3
• S B 11 O a> i^ O -j
't^o (Q S u N S o
^»E5i o+tSS*ot«S
R 1
E
2 ^^ ^
-H cN co Os
u . . x^ .
v ^ s - s 1 1
g-co 5co ^-_ _ p^Jr ^
^ g"^ S-oOs_°o at
• -3 1^ 2 SlJ J
^^-.^ ^^ a''5^ ° ""^^ «
B^^-^'S ^H»O B
o> »n os CN o^ ^ ^
•< co so — 55
< < <
Z Z Z
u b o £i o>
•1"^^ S'^S -a "^^.
S *S t-i jS *a ui •Si_
^^o^JTo ^J^0
UU 4) 4> 4> 1> U UUU
OO OO OO O OOO
ZZ ZZ £2; 2; K K K
1> CS ^O
0-^- OTf OO O 6 ^4) ««55^'2S5^W1)O
B "^ C* "^ C C M *" C t+ «••• •— « M
O ^ co co *»
c §
&s o & 1
S S ^ a 1 °°
S1 | « 5 S. •S
•J H -S >? M i4-c
tt c & « u o
""> oo t™~ O OS Os V} ^-^.
CO CO OO CN Vl CO O ^-
CN-** -• CN — ^-O
ooo o oooS
^;CC C dCd^:
^''t^ cN vO'»*«- "^. ^ Ti ^ '^. 'Ti to O
e Z, /-. Z Z, Z Z c c
O [I] tU PQ U L*J [JJ O O
Z Q Q Q Q Q Q Z Z
t** CS ^ ^O
OOO O OfN^OO *O
Cfl -_ Wl ._, .. ,. CO
S 4> 3 v)Mtf>3 to
«tf>rt 333« =
J3 OJ3_.ed rt «J3-^«
XT3 X^jXJSJS XuJ*;
k.k*Ur.O(XxxX4>MK
« '= -s 1 1 •§ S 2 S i S S
2 Slil-aiil s-ai
CuSsSe ?#?££•?
CN* 3 — CN — •* CO
oo o oooo o
U," tL," U." U- tt." ttJ 0.' li," U.'
ri"T" T "T*T"t""r "1"
SS2 S SSS2 S
lli
December 1994
7-5
DRAFT-DO NOT QUOTE OR CITE
-------
CO
I
p
CO
^
H
U
O
S
u
0«
<
J
1
"^
O
1^7
w
"^1
^5
^^
^J
CO
.
?
o
u
ti.
W
J
M
H
a
rs>
| 0
^ 0
II
(3 £
c
l_ U
u c
sS
H
q
.0
o 5 -^ <» -^ .
^ -4>'u "4>*aj B8*<
x<2 s'-^S x_Su x*S
rt^cxo 13"^^ -0"^^ •S"^SH
ItSs s|"u ^t°°:al>§'i
•^^feo o^fe o^S o^S6
voincisr) csi^c: tNf~<2 Sr-->2t--
1
SS S 8 1 SSSlsIsg
Iz SSI Illlllll
> D 5
D
<*! •* •* •* •* Tj- Tf
ooo o^osd ovboosdsoo^o
B ^ S & «5 « to
ii ill 1 1 1 1 S ! 1 1
«| S | | S|§8.§|ls
u? u## u?0u??o^
S °
o o in u: uT as
**. °! + + S s S s 9 5
^ S S S S u; u." u." u." S S
s < < <
i 1 S1 1 I s o
« n 0 2 •* 3 •*
M "2 fc O CJ O U
'w _: ^
*-• « m
u « T) oo
(O t. «J ^H
c B o .
— ^-v qj ^-^ -^ O
2? ^ o — • u S
2 oo ^ c» a G
t2c ISfeS
December 1994
7-6
DRAFT-DO NOT QUOTE OR CITE
-------
w
g
S
^
fc
u
2
"
o
o
5
u
3
U
ta
O
^
g
R
C/3
B
I
i
S
H
13
i
o
U
)£
n
CN O CN O ^H •""* OO Q\
ofioS OO (NO
1
| (OOOS-M S- rj £, — oo ~ .5
_ -^ ^ ~_- ^ ^ -^ ^v ^ ^ g
^ i 3 5 ^2. MSo^^o^,
E 0 0. 29- »g. a Bc.2-3 «
S'n^PSN'o'f-'!? ^^ g[ "^£0 51 "? "" en —
1 s d s d ci ci *
3 g (i
" S
M
| S-c?^f?^P
g id 2 =o 2 id ss-
"*
OO OO OO OO
|i J.
1? ilf!
u fi S t- £ 2
s^lsnl s a 1 a 1 Vs ssss
|g||gl 1 || || || Illl
ir>
cN t^ in
OOOfSCScN O O CNfN OfN O O O *^
V)V)tff MM M MMtA
S33 P3 3 S33
t, t, ts-a-sl u h II ..1 ,,111
•3 '3 '3 « u u '3 « ° u '3 u -g « o o
H«^OOO M PH OO _.O OOO
gggo'g'o a 3 !§!§ ""§ « "o "3 "3
000?^? 0 D ff 0? U#i^
S g
B." 11," 0000
S S SSSS
i§ i.i i
g g g 8 •*
^ u M G c4
•< W3 ^1 M eg
•o °°
S ^ | _B _
o ^ ^ co w en"
S. -S B, oo -S oo
O O ed O> ^ ON
o. o, -2 c. ? C
•a "
3 £
; inSTcn'S- O 5~f7
aco'mt^S S«2^
§ GC.aS5 c.2c
vooooo^X . oo^Tin
Cooooojg £mOOoo
gCCC?« «5C5
§o— M as
^
S 8
I j?
•3
o
3. °°
December 1994
7-7
DRAFT-DO NOT QUOTE OR CITE
-------
W5
H
j^
u^
g
>
H
U
O
o
U
«
U
_]
^
|
^
fa
O
^.
^\
3
U
^j
t/3
^
O
'"T*
TABLE 7
a
c
o
U
«
U
U
"2
"o
rt
8.
£
o
1
E c
3 O
a'f
1 1
<2 o
§0
II
x £
UJ cu
G
(- U
•5 *g
O u
£
c
Q
Particle
Concentre
(mg/m3
Exposure
Atmosphere
1
*J
U .S
'§ S
•o
o
J
^
05
|
'S
«
n
o
c
U
T3
<
-^ ^ 0
•n oo ^
005
^.^
22°-
005
005
§o o
E E
*O fsl OO
^
>; 8 °
* -a""
^o o
g g g
o o o
%**£,'£<
333
J3 JS J3
XXX
U U U
O 0 O
^ ? &
Ishinishi et al. Rat/F344 NS,
(1988b) NS,
NS,
002
*n ^ ^
0 0 £
2-2- 2-
o o g
005
ill
*O N oo
g g g
o o o
2 Z *2
333
X X *X
U U U
o o o
£j £j ij
CO CO CO
£
.5P
22°
^ jr ~
005
22°-
005
005
III
\o ts oo
_i<"
U
•§""
^H so 4S
S g §
o o o
Z Z Z
«-i in /•> o .5
a m 5
I 1 ** I
S M U C o
•a -r > 2 &
4* M « K
o 1 S °- U
— H O til *^l O
o e o 3
1 1 ^11
„ « ^b g -.
C P n-
^ « s § a S-
O cr* w *^ M |
§ <0 3 _ 3 »0
1 r^ g ^ .3
f ii? i 51 *
1 Ha I J| 1
I ££s ^ ^SS
1 ° o S M. S g JJ
S Mc/,J= ">u^"'C «
s p! |I £": S
e **a «la p'e.-i
o "i >a
I SSa. ISi slll
* « « * 5 g -a - 8 s K
"Table values indicate number exposed/number wi
bNumber of animals examined for tumors.
Significantly different from clean air controls.
dDiphenylnitrosamine; 6.25 mg/kg/week sc during
'Diphenylnitrasamine; 12.5 mg/kg/week sc during
'Splenic lymphomas also detected in controls (8.3!
E5.3% incidence of large cell carcinomas.
hl g/kg, ip I/week for 3 weeks starting 1 mo into
'Includes adenomas, squamous cell carcinomas, ad
J4.5 mg/DEN/kg, sc, 3 days prior to start of inhal
kSingle ip dose 1 mg/kg at start of exposure.
'Butylated hydro xy toluene 300 mg/kg, ip for Weel
m!2 mg/m^rom 12 weeks of age to termination of
"120-121 males and 71-72 females examined histol
°Not all animals were exposed for full term, at lea
7-8 DRAFT-DO NOT C
NS = Not specified.
NA = Not applicable.
)UO1
-------
1 certification cycles. The engines were equipped with automatic transmissions connected to
2 eddy-current dynamometers and flywheels simulating resistive and inertial loads of a midsize
3 passenger car. The D-2 diesel control fuel (Phillips Chemical Co.) met U.S. Environmental
4 Protection Agency (EPA) certification standards and contained approximately 30% aromatic
5 hydrocarbons and 0.3% sulfur. Following passage through a standard automotive muffler
6 and tail pipe, the exhaust was diluted 10:1 with filtered air in a dilution tunnel and serially
7 diluted to the final concentrations. The primary dilution process was such that particle
8 coagulation was retarded. Mokler et al. (1984) provided a detailed description of the
9 exposure system. The gas-phase components of the diesel exhaust atmospheres are presented
10 in Appendix A. No exposure-related changes in body weight or life span were noted for any
11 of the exposed animals nor were there any signs of overt toxicity. Collective lung tumor
12 incidence was greater (z statistic, p < 0.05) in the high (7.1-mg/m3) and medium
13 (3.5-mg/m3) exposure groups (12.8 and 3.6%, respectively) versus the control and low
14 (0.35-mg/m3) exposure groups (0.9 and 1.3%, respectively). Bronchoalveolar adenomas,
15 adenocarcinomas, and squamous cysts (considered benign, except for two that were classified
16 as squamous cell carcinomas because of the presence of less differentiated cells and invasion
17 of blood and lymph vessels) were identified. Using the same statistical analysis of specific
18 tumor types, adenocarcinoma plus squamous cell carcinoma and squamous cyst incidence was
19 significantly greater in the high-exposure group, and the incidence of adenomas was
20 significantly greater in the medium exposure group. A significant (p < 0.001)
21 exposure-response relationship was obtained for tumor incidence relative to exposure
22 concentration and lung burden of particulate matter (soot). These data are summarized in
23 Table 7-1. A logistic regression model estimating tumor prevalence as a function of time,
24 dose (lung burden of soot), and sex indicated a sharp increase in tumor prevalence for the
25 high dose level at about 800 days after the commencement of exposure. A less pronounced,
26 but definite, increase in prevalence with time was predicted for medium dose levels.
27 Significant effects were not detected at the low concentration. The particulate matter burdens
28 (mg per lung) of rats exposed to 0.35, 3.5, or 7.1 mg of soot/m3 for 24 mo were 0.6, 11.5,
29 and 20.8, respectively, and affirmed the greater than predicted accumulation that was the
30 result of decreased particle clearance following high-exposure conditions.
December 1994 7.9 DRAFT-DO NOT QUOTE OR CITE
-------
1 In summary, this study demonstrated the pulmonary carcinogenicity of high
2 concentrations of whole, diluted diesel exhaust in rats following chronic inhalation exposure.
3 Additionally, increasing lung particle burden resulting from this high-level exposure from a
4 decreased clearance was demonstrated. A logistic regression model presented by Mauderly
5 et al. (1987) indicated that both lung paniculate matter burden and exposure concentration
6 may be useful for expressing exposure-effect relationships.
7 A series of studies was conducted at the Fraunhofer-Institute in which female SPF
8 Wistar rats were exposed for 19 h/day, 5 days/week to both filtered and unfiltered (total)
9 diesel exhaust at an average particulate matter concentration of 4.24 mg/m3. Animals were
10 exposed for a maximum of 2.5 years. The exposure system as described by Heinrich et al.
11 (1986b), used a 40 kw 1.6-L diesel engine operated continuously under the U.S. 72 FTP
12 driving cycle. The engines used European Reference Fuel with a sulfur content of 0.36%.
13 Filtered exhaust was obtained by passing engine exhaust through a Luwa FP-65 HT
14 610 particle filter heated to 80 °C and a secondary series of filters (Luwa FP-85, Luwa
15 NS-30, and Drager CH 63302) at room temperature. The filtered and unfiltered exhausts
16 were each diluted 1:17 with filtered air and passed through respective 12-m3 exposure
17 chambers. Mass median aerodynamic diameter of the diesel exhaust particulate matter was
18 0.35 ± 0.10 /mi (mean + SD). The gas-phase components of the diesel exhaust
19 atmospheres are presented in Appendix A.
20 The effects of exposure to either filtered or unfiltered exhaust were described by
21 Heinrich et al. (1986a) and Stober (1986). Exposure to unfiltered exhaust resulted in eight
22 bronchoalveolar adenomas and nine squamous cell tumors in 15 of 95 rats examined for a
23 15.8% tumor incidence. Although statistical analysis was not provided, the increase appears
24 to be highly significant. In addition to the bronchio-alveolar adenomas and squamous cell
25 tumors, there was a high incidence of bronchio-alveolar hyperplasia (99%) and metaplasia of
26 the bronchio-alveolar epithelium (65%). No tumors were reported among female Wistar rats
27 exposed to filtered exhaust (n = 92) or clean air (n = 96).
28 Mohr et al. (1986) provided a more detailed description of the lung lesions and tumors
29 identified in Heinrich et al. (1986a,b) and Stober (1986). Substantial alveolar deposition of
30 carbonaceous particles was noted for rats exposed to the unfiltered diesel exhaust. Squamous
31 metaplasia was observed in 65.3% of the rats breathing unfiltered diesel exhaust but not in
December 1994 7_10 DRAFT-DO NOT QUOTE OR CITE
-------
1 any of the control rats. Of the nine squamous cell tumors, one was characterized as a
2 Grade I carcinoma (borderline atypia, few to moderate mitoses, and slight evidence of
3 stromal invasion), whereas the remaining eight were classified as benign, keratinizing, cystic
4 tumors.
5 The effect of chronic (19 h/day, 5 days/week, 2 to 2.5 years) diesel exhaust exposure
6 on the tumor-inducing effect of dipentylnitrosamine (DPN) was examined using female
7 Wistar rats (Heinrich et al., 1986a; Stober, 1986; Heinrich et al., 1989a). Groups of rats
8 (45 to 48 per group) were exposed to clean air or whole diesel exhaust (particle concentration
9 of 4.24 mg/m3, as described previously) and administered by subcutaneous injection 250 or
10 500 mg DPN/kg/week during the first 25 weeks of exposure. The total DPN dose
11 administered equalled 6.25 or 12.5 g/kg of body weight. The concentrations of B[0]P,
12 benzo[e]pyrene (B[e]P), and chrysene in the diesel exhaust were 13, 21, and 76 ng/m3,
13 respectively.
14 The overall tumor rate in the lungs of DPN-treated rats was not affected by the
15 exposure to either filtered or whole diesel engine exhaust. However, when only pulmonary
16 squamous cell carcinomas were considered, the exposure to whole diesel exhaust significantly
17 (p < 0.05) increased the tumor incidence (Table 7-1). Conversely, the high level of nasal
18 tumors induced by DPN was significantly decreased in the rats exposed to the diesel engine
19 emissions.
20 Heinrich et al. (1986b) and Mohr et al. (1986) compared the effects of exposure to
21 particles having only a minimal carbon core but a much greater concentration of PAHs than
22 does diesel paniculate matter. The desired exposure conditions were achieved by mixing coal
23 oven flue gas with pyrolyzed pitch. The concentration of B[fl]P and other PAHs/mg
24 paniculate matter was about three orders of magnitude greater than that of diesel exhaust.
25 Female rats were exposed to the flue gas-pyrolyzed pitch for 16 h/day, 5 days/week at
26 particle concentrations of 3 to 7 mg/m3 for 22 mo, then held in clean air for up to an
27 additional 12 mo. Of 116 animals exposed, 22 tumors were reported in 21 animals, for an
28 incidence of 18.1%. One was a bronchiolo-alveolar adenoma, one was a bronchiolo-alveolar
29 carcinoma, and 20 were squamous cell tumors. Among the latter, 16 were classified as
30 benign keratinizing cystic rumors and four were classified as carcinomas. No tumors were
December 1994 7-11 DRAFT-DO NOT QUOTE OR CITE
-------
1 reported in 115 controls. The tumor incidence in this study was comparable to that reported
2 previously for the diesel-exhaust-exposed animals.
3 In a recent, yet unpublished study, female Wistar rats were exposed for 18 h/day,
4 5 days/week for 10 or 20 mo to a carbon black (CB) aerosol at a mean concentration of
5 6.09 mg/m3 (Heinrich, 1990). The purpose of this study was to compare the carcinogenicity
6 of particles having no more than traces of PAHs with those of diesel exhaust. Groups of
7 72 animals each were exposed for either 10 mo and held an additional 20 mo in clean air or
8 for 20 mo then held an additional 10 mo in clean air. Among the rats exposed for 10 mo,
9 malignant tumors were seen in 14% and benign tumors in additional 3%. These included
10 three animals with bronchio-alveolar adenocarcinoma, seven with squamous-cell carcinoma,
11 one with an adenosquamous carcinoma and two with bronchiolo-alveolar adenoma. In the
12 group exposed for 20 mo, the tumor incidence was 8%, but all were malignant (one animal
13 with bronchiolo-alveolar adenocarcinoma and five with squamous-cell carcinoma. No tumors
14 were seen in the control groups. Another study in which the comparative tumorigenicity of
15 CB and diesel exhaust was conducted at the Inhalation Toxicology Research Institute
16 (Mauderly et al., 1991; Nikula et al., 1991, 1994). In this study, F344 rats were exposed to
17 either CB or diesel exhaust for 16 h/day, 5 days/week to particle concentrations of 2.5 or
18 6.5 mg/m3 for up to 24 mo. Controls were exposed to clean air. Preliminary results show
19 that both diesel exhaust and carbon black are pulmonary carcinogens under the exposure
20 conditions of the study (Nikula et al., 1991). Nineteen animals in both the high-exposure
21 diesel exhaust (HDE) and the high-exposure carbon black (HCB) groups exhibited primary
22 lung tumors (grossly observed or suspected and histologically confirmed). For the HDE
23 group there were 8 squamous cysts, 4 adenomas, 37 adenocarcinomas, and 2 squamous cell
24 carcinomas, and for the HCB group there were 17 squamous cysts, 6 adenomas,
25 26 adenocarcinomas, 2 squamous cell carcinomas, and 1 "other" malignant tumor. Primary
26 lung tumors were observed in 11 rats of the low-exposure diesel exhaust (LDE) group and
27 four rats of the low exposure carbon black (LCB) group. For the LDE group there were
28 3 squamous cell cysts, 2 adenomas, 7 adenocarcinomas, and 3 squamous cell carcinomas,
29 and for the LCB group there were 7 squamous cell cysts, 1 adenocarcinoma, 2 squamous cell
30 carcinomas, and 1 "other" malignant tumor.
December 1994 7-12 DRAFT-DO NOT QUOTE OR CITE
-------
1 In analyzing the studies of Heinrich et al. (1986a,b), Heinrich (1990), Mohr et al.
2 (1986), and Stober (1986), it must be noted that the incidence of lung tumors occurring
3 following exposure to whole diesel exhaust, coal oven flue gas, or CB (15.8, 18.1, and 8 to
4 17%, respectively) was very similar. This occurred despite the fact that the PAH content of
5 the PAH-enriched pyrolyzed pitch was more than three orders of magnitude greater than that
6 of diesel exhaust; CB on the other hand, had only traces of PAHs. Based on these findings,
7 the organic fraction is not the sole cause of tumor induction by diesel exhaust. This issue is
8 discussed further in Chapter 10.
9 A long-term inhalation study (Ishinishi et al., 1988b; Takaki et al., 1989) examined the
10 effects of emissions from light-duty (LD) and heavy-duty (HD) diesel engines on male and
11 female Fischer 344/Jcl rats. The LD engines were 1.8-L, 4-cylinder, swirl-chamber-type
12 power plants, and the HD engines were 11-L, 6-cylinder, direct-injection-type power plants.
13 The engines were connected to eddy-current dynamometers and operated at 1,200 rpm (LD
14 engines) and 1,700 rpm (HD engines). Nippon Oil Co. JIS No. 1 or No. 2 diesel fuel was
15 used. The 30-mo whole-body exposure protocol (16 h/day, 6 days/week) employed diesel
16 exhaust particle concentrations of 0, 0.5, 1, 1.8, or 3.7 mg/m3 from HD engines and 0, 0.1,
17 0.4, 1.1, or 2.3 mg/m3 from LD engines. The B[a]P concentrations were reported as
18 4.4 and 2.8 /ig/g of paniculate matter, and 1-nitropyrene concentrations were 57.1 and
19 15.3 jug/g of paniculate matter for the LD and HD engines, respectively. An analysis of
20 gas-phase components is presented in Appendix A. The animals inhaled the exhaust
21 emissions from 1700 to 0900 hours. Sixty-four male rats and 59 to 61 female rats from each
22 exposure group were evaluated for carcinogenicity.
23 For the experiments using the LD series engines, the highest incidence of hyperplastic
24 lesions plus tumors (72.6%) was seen in the highest exposure (2.3 mg/m3) group. However,
25 this high value was the result of the 70% incidence of hyperplastic lesions; the incidence of
26 adenomas was only 0.8% and that of carcinomas 1.6%. Hyperplastic lesion incidence was
27 considerably lower for the lower-exposure groups (9.7, 4.8, 3.3, and 3.3% for the 1.1-,
28 0.4-, and 0.1-mg/m3 and control groups, respectively). The incidence of adenomas and
29 carcinomas, combining males and females, was not significantly different among exposure
30 groups (2.4, 4.0, 0.8, 2.4, and 3.3% for the 2.3-, 1.1-, 0.4-, and 0.1-mg/m3 groups and the
31 controls, respectively).
December 1994 7_13 DRAFT-DO NOT QUOTE OR CITE
-------
1 For the experiments employing the HD series engines, the total incidence of
2 hyperplastic lesions, adenomas, and carcinomas was highest (26.6%) in the 3.7-mg/m3
3 exposure group. The incidence of adenomas plus carcinomas for males and females
4 combined equalled 6.5, 3.3, 0, 0.8, and 0.8% at 3.7, 1.8, 1, and 0.4 mg/m3 and for
5 controls, respectively. A statistically significant difference was reported between the
6 3.7 mg/m3 and the control groups for the HD series engines. A progressive dose-response
7 relationship was not demonstrated. Tumor incidence data for this experiment are presented
8 in Table 7-1.
9 The Ishinishi et al. (1988b) study also included recovery tests in which rats exposed to
10 whole diesel exhaust (particle concentration of 0.1 or 1.1 mg/m3 for the LD engine and
11 0.5 or 1.8 mg/m3 for the HD engine) for 12 mo were examined for lung tumors following
12 6-, 12-, or 18-mo recovery periods in clean air. The incidences of neoplastic lesions were
13 low, and pulmonary soot burden was lower than for animals continuously exposed to whole
14 diesel exhaust and not provided a recovery period. The only carcinoma observed was in a
15 rat examined 12 mo following exposure to exhaust (1.8 mg/m3) from the HD engine.
16 Iwai et al. (1986) also examined the long-term effects of diesel exhaust inhalation on
17 female F344 rats. The exhaust was generated by a 2.4-L displacement truck engine. The
18 exhaust was diluted 10:1 with clean air at 20 to 25 °C and 50% relative humidity. The
19 engines were operated at 1,000 rpm with an 80% engine load. These operating conditions
20 were found to produce exhaust with the highest particle concentration and lowest NO2 and
21 SO2 content. For those chambers using filtered exhaust, proximally installed high-efficiency
22 paniculate air (HEPA) filters were employed. Three groups of 24 rats each were exposed to
23 unfiltered diesel exhaust, filtered diesel exhaust, or filtered room air for 8 h/day,
24 7 days/week for 24 mo. Particle concentration was 4.9 mg/m3 for unfiltered exhaust.
25 Concentrations of gas-phase exhaust components were 30.9 ppm NOX, 1.8 ppm N02,
26 13.1 ppm SO2, and 7.0 ppm CO.
27 No lung tumors were found in the 2-year control (filtered room air) rats, although one
28 adenoma was noted in a 30-mo control rat, providing a spontaneous tumor incidence of
29 4.5%. No lung tumors were observed for rats exposed to filtered diesel exhaust. Four of
30 14 rats exposed to unfiltered diesel exhaust for 2 years developed lung tumors, two of these
31 being malignant. Five rats of this 2-year exposure group were subsequently placed in clean
December 1994 7-14 DRAFT-DO NOT QUOTE OR CITE
-------
1 room air for 3 to 6 mo and four eventually (time not specified) exhibited lung tumors (three
2 malignancies). Thus, the lung tumor incidence for total tumors was 42.1% (8/19) and
3 26.3% (5/19) for malignant tumors in rats exposed to whole diesel exhaust. The tumor types
4 identified were adenomas (3/19), adenocarcinomas (1/19), adenosquamous carcinoma (2/19),
5 squamous carcinoma (1/19), and large-cell carcinoma (1/19). The lung tumor incidence in
6 rats exposed to whole diesel exhaust was significantly greater than that of controls
7 (p < 0.01). Tumor data are summarized in Table 7-1. Malignant splenic lymphomas were
8 detected in 37.5% of the rats in the filtered exhaust group and in 25.0% of the rats in the
9 unfiltered exhaust group, these values being significantly (p < 0.05) greater than the 8.2%
10 incidence noted for the control rats. The study demonstrates production of lung cancer in
11 rats following 2-year exposure to unfiltered diesel exhaust. Additionally, the occurrence of
12 splenic, malignant lymphomas occurred during exposure to both filtered and unfiltered diesel
13 exhaust. This is the only report to date of tumor induction at an extrarespiratory site.
14 A chronic (up to 24 mo) inhalation exposure study by Takemoto et al. (1986) was
15 conducted to determine the effects of diesel exhaust, di-isopropanol-nitrosamine (DIPN), and
16 diesel exhaust following DIPN treatment to female F344/Jcl rats. One month after initiation
17 of inhalation exposures, DIPN was administered ip at 1 mg/kg weekly for 3 weeks to clean
18 air and diesel exposed groups of rats. Uninjected groups were also exposed to clean air and
19 diesel exhaust. The treatment protocol consisted of exposure to diesel exhaust for 4 h/day,
20 4 days/week. The diesel exhaust was generated by a 269-cc displacement engine operated at
21 an idle state (1,600 rpm). Concentrations of the gas-phase components of the exhaust are
22 presented in Appendix A. The particle concentration of the diesel exhaust in the exposure
23 chamber was 2 to 4 mg/m3. Benzo[a]pyrene and 1-nitropyrene concentrations were 0.85 and
24 93 fig/g of particles, respectively.
25 In the Takemoto et al. (1986) study, no lung tumors were reported in either uninjected
26 controls or diesel-exposed animals. Among injected animals autopsied at 12 to 17 mo,
27 2 adenomas were reported in 8 rats exposed to clean air compared with 12 adenomas and
28 3 adenocarcinomas in 18 diesel-exposed rats. Among injected rats autopsied at 18 to 24 mo,
29 10 adenomas and 4 adenocarcinomas were seen in 21 animals exposed to clean air compared
30 with 12 adenomas and 7 adenocarcinomas in 18 diesel-exposed rats. According to the
31 authors, the incidence of malignant tumors was not significantly increased in either of the
December 1994 745 DRAFT-DO NOT QUOTE OR CITE
-------
1 diesel exhaust-exposed groups when compared with the appropriate control group. Tumor
2 incidence data for the various treatment protocols are presented in Table 7-1. It was also
3 noted that the diesel engine employed in this study was originally used as an electrical
4 generator and that its operating characteristics (not specified) were different from those for a
5 diesel-powered automobile; however, the investigators deemed it suitable for assessing the of
6 effects diesel emissions.
7 Brightwell et al. (1986, 1989) studied the effects of filtered and unfiltered diesel
8 exhaust on male and female F344 rats. The diesel exhaust was generated by a 1.5-L
9 Volkswagen engine that was computer-operated according to the U.S. 72 FTP driving cycle.
10 The engine emissions were diluted by conditioned air delivered at 800 m3/h to produce the
11 high-exposure (6.6 mg/m3) diesel exhaust atmosphere. Further dilutions of 1:3 and
12 1:9 produced the medium- (2.2 mg/m3) and low- (0.7 mg/m3) exposure atmospheres.
13 Filtered diesel exhaust was generated by a similar engine. The CO and NOX concentrations
14 (mean ± SD) were 32 ± 11 ppm and 8 ± 1 ppm for the unfiltered diesel exhaust
15 (high-exposure concentration chamber) and 32 ± 11 and 8 ± 1 for the filtered diesel
16 exhaust. The inhalation exposures were conducted overnight to provide five 16-h periods per
17 week for 2 years; surviving animals were maintained for an additional 6 mo.
18 For males and females combined, a 9.7% (14/144) and 38.5% (55/143) incidence of
19 primary lung tumors occurred in F344 rats following exposure to 2.2 and 6.6 mg of diesel
20 soot/m3, respectively (Table 7-1). The tumor incidence in the 0.7-mg/m3 exposure group
21 was 0.7% (1/144) and that of controls was 1.2% (3/260). Diesel-exhaust-induced tumor
22 incidence in rats was dose-related and higher in females than males (Table 7-1). These data
23 included animals sacrificed at the interim periods (6, 12, 18, and 24 mo) therefore, the tumor
24 incidence does not accurately reflect the effects of long-term exposure to the diesel exhaust
25 atmospheres. When tumor incidence is expressed relative to the specific interim sacrifice
26 period, a lung tumor incidence of 96% (24/25), 76% (19/25) of which were malignant, was
27 reported for female rats in the high-dose group exposed for 24 mo and held in clean air for
28 the remainder of their lives. For male rats in the same group, the tumor incidence equalled
29 44% (12/27), of which 37% (10/27) were malignant. It was also noted that many of the
30 animals exhibiting tumors had more than one tumor, often representing multiple histological
31 types. The types of tumors identified in the rats exposed to diesel exhaust included
December 1994 7_16 DRAFT-DO NOT QUOTE OR CITE
-------
1 adenomas, squamous cell carcinomas, adenocarcinomas, mixed adenoma/adenocarcinomas,
2 and mesotheliomas. Similar to other studies, the tumor incidence in rats occurred during
3 exposure to whole exhaust rather than filtered exhaust. It must be noted, however, that the
4 exposure during darkness (when increased activity would result in greater respiratory
5 exchange and greater inhaled dose) could account, in part, for the high response reported for
6 the rats.
7 Karagianes et al. (1981) exposed male Wistar rats (40 per group) to diesel engine
8 exhaust diluted to a particle concentration of 8.3 (± 2.0) mg/m3, room air, diesel engine
9 exhaust (8.3 mg/m3) plus low-concentration coal dust (5.8 mg/m3), low-concentration coal
10 dust only (6.6 mg/m3), or high-concentration coal dust (14.9 mg/m3) 6 h/day, 5 days/week
11 for up to 20 mo. The exhaust-generating system and exposure atmosphere characteristics are
12 presented in Appendix A. The type of engine used (3-cylinder, 43-bhp diesel) is normally
13 used in mining situations and was connected to an electric generator and operated at varying
14 loads and speeds to simulate operating conditions in an occupational situation. To control the
15 CO concentration at 50 ppm, the exhaust was diluted 35:1 with compressed air.
16 One bronchiolar adenoma was detected in the group exposed to diesel exhaust alone
17 and one in the rats receiving combined exposures. No lung tumors were reported in controls
18 or following exposure to either high or low concentrations of coal dust. The equivocal
19 response may have been caused by the relatively short exposure durations (20 mo). In the
20 Mauderly et al. (1987) study, by comparison, most of the tumors were detected in rats
21 exposed for more than 24 mo.
22 Lewis et al. (1986) also examined the effects of inhalation exposure of diesel exhaust
23 and/or coal dust on tumorigenesis on F344 rats. Groups of 216 male and 72 female rats
24 were exposed to clean air, whole diesel exhaust (2 mg soot/m3), coal dust (2 mg/m3), or
25 diesel exhaust plus coal dust (1 mg/m3 of each) for 7 h/day, 5 days/week for up to 24 mo.
26 Groups of 10 or more males were sacrificed at intermediate intervals (3, 6, and 12 mo). The
27 diesel exhaust was produced by a 7.0-L, four-cycle, water-cooled Caterpillar Model 3304
28 engine using No. 2 diesel fuel (<0.5% sulfur by mass). The exhaust was passed through a
29 Wagner water scrubber, which lowered the exhaust temperature and quenched engine
30 backfire. An analysis of the exposure atmospheres is presented in Appendix A.
December 1994 7_17 DRAFT-DO NOT QUOTE OR CITE
-------
1 Histological examination was performed on 120 to 121 male and 71 to 72 female rats
2 terminated after 24 mo of exposure. No specific tumor information was provided other than
3 that the exhaust exposure did not significantly affect the tumor incidence. There was no
4 postexposure period, which may explain, in part, the lack of significant tumor induction.
5 The paniculate matter concentration was also less than the effective dose in several of the
6 other studies.
7 General Motors Research Laboratories sponsored chronic inhalation studies using male
8 Fischer 344 rats exposed to diesel exhaust particle concentrations of 0.25, 0.75, or
9 1.5 mg/m3 (Kaplan et al., 1983; White et al., 1983). The exposure protocol for this study
10 conducted at the Southwest Research Institute (SWRI) was 20 h/day, 7 days/week for 9 to
11 15 mo. Some animals were sacrificed following completion of exposure, whereas others
12 were returned to clean-air atmospheres for an additional 8 mo. Control animals received
13 clean air. Exhaust was generated by 5.7-L Oldsmobile engines (four different engines used
14 throughout the experiment) operated at a steady speed and load simulating a 40-mph driving
15 speed of a full-size passenger car. Details of the exhaust-generating system and exposure
16 atmosphere are presented in Appendix A.
17 Five instances of bronchoalveolar carcinoma were observed out of 90 rats exposed to
18 diesel exhaust for 15 mo and held an additional 8 mo in clean air. These included one tumor
19 in the 0.25-mg/m3 group, three in the 0.75-mg/m3 group, and one in the 1.5-mg/m3 group.
20 Rats kept in clean air chambers for 23 mo did not exhibit any carcinomas. No tumors were
21 observed in any of the 180 rats exposed to diesel exhaust for 9 or 15 mo without a recovery
22 period or in the respective controls for these groups. Although the increase in tumor
23 incidences in the groups exposed for 15 mo and held an additional 8 mo in clean air were not
24 statistically significant, they are suggestive of an effect because the background incidence for
25 this specific lesion in this strain of rat is low.
26
27 7.2.2 Mouse Studies
28 Heinrich et al. (1986a) and Stober (1986), as part of a larger study, also evaluated the
29 effects of diesel exhaust in mice. Details of the exposure conditions reported by Heinrich
30 et al. (1986b) are given in Appendix A. Following lifetime (19 h/day, 5 days/week, for a
31 maximum of 120 weeks) exposure to filtered (n = 93) and unfiltered (n = 76) diesel exhaust
December 1994 7-18 DRAFT-DO NOT QUOTE OR CITE
-------
1 (4.2.4 mg/m3), female NMRI mice exhibited a total lung tumor incidence of adenomas and
2 adenocarcinomas combined of 31% (filtered) and 32% (unfiltered), respectively. Tumor
3 incidences reported for control mice (n = 84) equalled 11 % for adenomas and
4 adenocarcinomas combined. The effects are more dramatic when the incidences of only
5 malignant tumors (adenocarcinomas) are considered, 2.4% for controls, 19% for filtered
6 exhaust, and 17% for unfiltered exhaust. This is the only reported study in which filtered
7 exhaust resulted in a definitive tumorigenic response in the lungs of mice. These data are
8 summarized in Table 7-1.
9 As part of the same study, groups of 64 female NMRI mice of 8 to 10 weeks of age
10 were dosed weekly with either 50 or 100 /*g E[a]P intratracheally for 20 or 10 weeks,
11 respectively, for a total dose of 1 mg. Another group received 50 pig dibenz[fl,/z]-
12 anthracene (DBA) intratracheally for 10 weeks. Additional groups of 96 newborn mice
13 received one sc injection of 5 or 10 /*g of DBA between 24 and 48 h after birth. The
14 animals were concomitantly exposed to either diesel exhaust or clean air. The mice
15 receiving intratracheal instillations were observed throughout their lifespan but the newborn
16 mice were sacrificed after 6 mo. Although the chemical treatments resulted in large
17 increases in lung tumor incidence, exposure to diesel exhaust did not enhance this effect and
18 in some cases even resulted in inhibition. For example, lung tumor rates in clean air mice
19 treated with 20 instillations of B[a]P equalled 71% compared with 41% for mice similarly
20 instilled but exposed to diesel exhaust. The decrease resulted from a smaller number of
21 adenocarcinomas, whereas the adenoma incidence remained unchanged. The high dose of
22 DBA injected into newborn mice also resulted in a greater tumor incidence in mice exposed
23 to clean air (81 %) than in the diesel exposed group (63%). Effects of the other treatments
24 were apparently not inhibited by diesel exhaust exposure, although complete incidence data
25 were not reported. The authors did not speculate on the reasons for this unexpected effect.
26 Takemoto et al. (1986) reported the effects of inhaled diesel exhaust (2 to 4 mg/m3,
27 4 h/day, 4 days/week, for up to 28 mo) in ICR and C57BL mice exposed from birth.
28 Details of the exposure conditions are presented in Appendix A. Among male and female
29 ICR mice autopsied at 13 to 18 mo, 4 adenomas and 1 adenocarcinoma were detected in
30 34 diesel exhaust-exposed mice compared with 3 adenomas among 38 controls. Among
31 animals autopsied at 19 to 28 mo, 6 adenomas and 3 adenocarcinomas were seen in
December 1994 7_19 DRAFT-DO NOT QUOTE OR CITE
-------
1 22 exposed animals, compared with 3 adenomas and one adenocarcinoma in 22 controls.
2 Among combined male and female C57BL mice autopsied at 13 to 18 mo, 4 adenomas and
3 2 adenocarcinomas were detected in 79 animals autopsied compared with none in
4 19 unexposed animals. Among males and females autopsied at 19 to 28 mo, 8 adenomas and
5 3 adenocarcinomas were detected in 71 exposed animals compared with one adenoma among
6 32 controls. No significant increases in either adenoma or adenocarcinoma incidences were
7 reported for either strain of exposed mice. Although not tested by the authors, the combined
8 incidence of adenomas and adenocarcinomas (11/71) in male and female C57BL mice
9 exposed to diesel exhaust for 19 to 28 mo versus that found in controls (1/32), however,
10 appears to be a significant increase. Although the results are not definitive, there is the
11 strong suggestion of an effect, especially since the C57BL strain has a low background lung
12 tumor incidence. See Table 7-1 for details of tumor incidence.
13 Pepelko and Peirano (1983) summarized a series of studies on the health effects of
14 diesel emissions in mice. Exhaust was provided by two Nissan CN 6-33, 6-cylinder, 3.24-L
15 diesel engines coupled to a Chrysler A-272 automatic transmission and Eaton model 758-DG
16 dynamometer. Details of the exhaust generating system and the exposure atmosphere are
17 presented in Appendix A. Sixty-day pilot studies were conducted at a 1:14 dilution,
18 providing particle concentrations of 6 mg/m3- The engines were operated using the Modified
19 California Cycle. These 20-h/day, 7-day/week pilot studies using rats, cats, guinea pigs, and
20 mice produced decreases in weight gain and food consumption. Therefore, at the beginning
21 of the long-term studies, exposure time was reduced to 8 h/day, 7 days/week at an exhaust
22 particle concentration of 6 mg/m3. During the final 12 mo of exposure, however, the
23 particle concentration was increased to 12 mg/m3. For the chronic studies, the engines were
24 operated using the Federal Short Cycle.
25 Pepelko and Peirano (1983) described a two-generation study using Sencar mice
26 exposed to diesel exhaust alone or treated with either tumor initiators or promoters. Male
27 and female parent generation mice were exposed to diesel exhaust at a particle concentration
28 of 6 mg/m3 prior to (from weaning to sexual maturity) and throughout mating. The dams
29 continued exposure through gestation, birth, and weaning. Groups of offspring (130 males
30 and 130 females) received ip injections of either butylated hydroxytoluene (BHT) (300 mg/kg
31 for week 1, 83 mg/kg for week 2, and 150 mg/kg from week 3 to 1 year), a single ip
December 1994 7-20 DRAFT-DO NOT QUOTE OR CITE
-------
1 injection of 1 mg urethan at 6 weeks of age, or no injections. The exhaust exposure was
2 increased to a particle concentration of 12 mg/m3 when the offspring were 12 weeks of age
3 and was maintained until termination of the experiment when the mice were 15 mo old.
4 The incidence of pulmonary adenomas (16.3%) was significantly increased in the
5 noninjected female mice exposed to diesel exhaust, compared with 6.3% in clean air
6 controls. The incidence in males and females combined was 10.2% in 205 animals examined
7 compared with 5.1% in 205 clean air controls. This difference was also significant. The
8 incidence of carcinomas was not affected by exhaust exposure in either sex. Exhaust
9 exposure reduced the adenoma incidence in female mice receiving BHT (3.9 versus 16.7%).
10 The response to BHT in males, or urethan in both sexes was unaffected by diesel exposure.
11 These results provided the earliest evidence for cancer induction following inhalation
12 exposure to diesel exhaust. The limited response may well have been influenced by the
13 relatively early sacrifice times of the mice. On the other hand, an increase in the sensitivity
14 of the study, allowing detection of tumors at 15 mo, may have been the result of exposure
15 from conception. It is interesting to note that in this study diesel exposure appeared to inhibit
16 effects of tumor promotion, whereas Stober (1986) reported diesel exposure inhibition of
17 complete carcinogens. These data are summarized in Table 7-1.
18 A series of inhalation studies, using strain A mice, was conducted by Orthoefer et al.
19 (1981). In assays with the strain A, mice are usually given a series of sc injections with the
20 test agent; they are then sacrificed at about 9 mo of age and examined for lung tumors.
21 In the present series inhalation exposure was substituted. In the current series, groups of
22 25 male Strong A mice were exposed to irradiated (to simulate chemical reactions induced by
23 sunlight) or nonirradiated diesel exhaust (6 mg/m3) for 20 h/day, 7 days/week for 7 weeks.
24 Additional groups of 40 Jackson A (20 of each sex) were exposed similarly to either clean air
25 or diesel exhaust then held in clean air until sacrificed at 9 mo of age. No tumorigenic
26 effects were detected at nine months of age. Further studies were conducted in which male
27 A/Strong mice were exposed 8 h/day, 7 days/week until sacrifice (approximately 300 at 9 mo
28 of age and approximately 100 at 12 mo of age). With the exception of those treated with
29 urethan, the number of tumors per mouse did not exceed historical control levels in any of
30 the studies. Exposure to diesel exhaust, however, significantly inhibited the tumorigenic
31 effects of the 5-mg urethan treatment. Results are listed in Table 7-1.
December 1994 7_2i DRAFT-DO NOT QUOTE OR CITE
-------
1 Kaplan et al. (1982) also reported the effects of diesel exposure in strain A mice.
2 Groups of male strain A/J mice were exposed for 20 h/day, 7 days/week for 90 days, and
3 held until 9 mo of age. Experimental conditions are described in Appendix A. Briefly, the
4 animals were exposed to diesel exhaust at particle concentrations of 0, 0.25, 0.75, or
5 1.5 mg/m3. Controls were exposed to clean air. Among 458 controls and 485 exposed
6 animals, tumors were detected in 31.4% of those breathing clean air versus 34.2% of those
7 exposed to diesel exhaust. The mean number of tumors per mouse also failed to show
8 significant differences.
9 In a follow-up study, strain A mice were exposed to diesel exhaust for 8 mo (Kaplan
10 et al., 1983; White et al., 1983). After exposure to the highest exhaust concentration
11 (1.5 mg/m3), the percentage of mice with pulmonary adenomas and the mean number of
12 tumors per mouse were significantly less (p < 0.05) than for controls (25.0 versus 33.5%
13 and 0.30 ± 0.02 [S.E.] versus 0.42 ± 0.03 [S.E.]) (Table 7-1).
14
15 7.2.3 Hamster Studies
16 Heinrich et al. (1982) examined the effects of diesel exhaust exposure on the tumor
17 frequency in female Syrian golden hamsters pretreated with the tumor initiators DBA or
18 diethylnitrosamine (DEN). At the time of this work, it was presumed that traditional
19 inhalation exposure experiments would not result in definitive tumor formation; thus a tumor
20 initiation animal model was used. Groups of 48 to 72 animals were exposed to clean air,
21 whole diesel exhaust at a mean particle concentration of 3.9 mg/m3, or filtered diesel exhaust
22 with either no further treatment or administered DBA (intratracheal instillations of
23 0.1 mg/week for 20 weeks), DEN (1.5 or 4.5 mg/kg sc) or pyrene (intratracheal instillations
24 of 0.1 mg/week for 20 weeks), the last serving as a noncarcinogenic PAH control.
25 Inhalation exposures were conducted 7 to 8 h/day, 5 days/week for 2 years. The exhaust
26 was produced by a 2.4-L Daimler-Benz engine operated at 2,400 rpm.
27 Only two hamsters exhibited lung tumors, both having died during the exposure period.
28 One occurred in a hamster receiving DBA and exposed to filtered diesel exhaust for
29 75 weeks; the other occurred in a hamster receiving DEN and exposed to whole diesel
30 exhaust for 67 weeks. Compared with corresponding treatment groups, there was a higher
31 incidence of adenomatous proliferative changes in the lungs of hamsters exposed to whole
December 1994 7-22 DRAFT-DO NOT QUOTE OR CITE
-------
1 diesel exhaust. Hamsters exposed to filtered diesel exhaust also showed a greater incidence
2 of adenomatous proliferative changes than did those of the respective clean air exposure
3 groups. The incidence of proliferative changes in the lungs of hamsters receiving DEN or
4 DBA was greater than for those groups not treated with the initiators. Although not
5 definitive, this study provided information suggesting the possible involvement of whole
6 diesel exhaust and filtered diesel exhaust in producing histologic alterations in the lungs of
7 hamsters, though no increases in tumors were observed.
8 In a more recent study, Syrian hamsters were exposed 19 h/day, 5 days/week for a
9 lifetime to diesel exhaust diluted to a particulate matter concentration of 4.24 mg/m3
10 (Heinrichet al., 1986a; Stober, 1986). Details of the exposure conditions are reported in
11 Appendix A. Ninety-six animals per group were exposed to clean air, whole exhaust, or
12 filtered exhaust. Additional groups were treated with DEN (4.5 mg/kg, sc) or B[a]P
13 (20 doses of 0.25 mg intratracheal instillation) and exposed to the three experimental
14 atmospheres. No lung tumors were seen in uninjected clean-air or in either diesel
15 exhaust-exposed group. Initial treatment with DEN or B[a]P resulted in lung tumor
16 incidences of only 10 and 2%, respectively, which were not significantly changed by
17 exposure to diesel exhaust.
18 Heinrich et al. (1989b) reported results of experiments assessing the effects of DEN
19 and diesel exhaust exposure in combination. Hamsters were exposed to exhaust from a
20 Daimler-Benz 2.4-L engine operated at a constant load of about 15 kW and at a uniform
21 speed of 2,000 rpm. The exhaust was diluted to an exhaust-clean-air ratio of about
22 1:13, resulting in a mean particle concentration of 3.75 mg/m3. The animals were exposed
23 19 h/day, 5 days/week beginning at noon each day, under a 12-h light cycle, starting at
24 0700 hours. DEN (3 or 6 mg/kg) was given as a single sc injection 2 weeks from the start
25 of exposure to groups of 40 male and 40 female Syrian golden hamsters exposed to whole
26 diesel exhaust, filtered diesel exhaust, or clean air. Groups were also exposed to the exhaust
27 without DEN or to only clean air. Exposures were conducted in chambers maintained at
28 22 to 24 °C and 40 to 60% relative humidity for up to 18 mo. Surviving hamsters were
29 maintained in clean air for up to an additional 6 mo. The concentrations of B[a]P and B[e]P
30 in the whole exhaust atmospheres were 37.5 and 61.9 ng/m3, respectively.
December 1994 7_23 DRAFT-DO NOT QUOTE OR CITE
-------
1 No lung tumors were detected in any of the hamsters of any of the treatment groups.
2 A nasal carcinoma was detected in a female hamster treated with DEN (6 mg/kg) and
3 exposed to filtered exhaust. A tracheal carcinoma was detected in a male hamster exposed to
4 whole diesel exhaust and receiving DEN (3 mg/kg), and a laryngeal carcinoma observed in a
5 male hamster receiving DEN (6 mg/kg) and exposed to whole diesel exhaust. Exposure of
6 male hamsters to whole or filtered diesel exhaust alone did not result in a significant increase
7 in the tumors relative to clean air controls. Male hamsters receiving 6 mg DEN/kg plus
8 whole diesel exhaust exposure and dying before or after the 50% survival date, however, did
9 show an increase in tumor rate compared with DEN-treated animals exposed to clean air.
10 Using life-table analysis, a significant (p < 0.05) exposure-related increase in tumor rate
11 was noted for this group (40.0 versus 7.0% for filtered exhaust + DEN and 7.0% for clean
12 air + DEN). No upper respiratory tract tumors were detected in clean-air controls, or
13 filtered-exhaust-exposed groups that did not receive the DEN treatment.
14 In summary, diesel exhaust alone did not induce an increase in lung tumors in hamsters
15 of either sex. Diesel exhaust did significantly enhance the tumorigenic effects of DEN in
16 males injected with 6 mg DEN/kg but not in females or in either sex given the 3 mg/kg
17 dose. The cocarcinogenic effects of diesel with DEN therefore appear to be equivocal.
18 Brightwell et al. (1986, 1989) studied the effects of filtered or unfiltered diesel exhaust
19 on male and female Syrian golden hamsters. Groups of 52 males and 52 females received no
20 injections or sc injections of 4.5 mg DEN/kg 3 days before the start of exposure. The
21 animals were 6 to 8 weeks old at the start of exposure to diesel exhaust at particle
22 concentrations of 0.7, 2.2, or 6.6 mg/m3. They were exposed 16 h/day, 5 days/week for a
23 total of 2 years and then sacrificed. Exposure conditions are described in Appendix A.
24 Although the DEN-pretreated hamsters exhibited an increase in tracheal papillomas in all
25 treatment groups when compared with non-DEN pretreated hamsters, there was no
26 statistically significant (t-test) relationship between tumor incidence and exhaust exposure.
27 As noted in International Agency for Research on Cancer (1989), however, the reporting of
28 tumor incidence and survival was incomplete.
29
30
December 1994 7_24 DRAFT-DO NOT QUOTE OR CITE
-------
1 7.2.4 Monkey Studies
2 Fifteen male cynomolgus monkeys were exposed to diesel exhaust (2 mg/m3) for
3 7 h/day, 5 days/week for 24 mo (Lewis et al., 1986). The same numbers of animals were
4 also exposed to coal dust (2 mg/m3), diesel exhaust plus coal dust (1 mg/m3 for each
5 component), or filtered air. Details of exposure conditions were listed previously in the
6 description of the Lewis et al. (1986) study with rats (Appendix A).
7 None of the monkeys exposed to diesel exhaust exhibited a significantly increased
8 incidence of preneoplastic or neoplastic lesions. It should be noted, however, that the 24-mo
9 time frame employed hi this study may not have allowed the manifestation of tumors in
10 primates, because this duration is only a small fraction of the monkeys expected life span.
11
12
13 7.3 LUNG IMPLANTATION OR INTRATRACHEAL INSTILLATION
14 STUDIES
15 7.3.1 Rat Studies
16 Grimmer et al. (1987), using female Osborne Mendel rats (35 per treatment group),
17 provided evidence that the PAHs in diesel exhaust that consist of four or more rings have a
18 carcinogenic potential. Condensate was obtained from the whole exhaust of a 3.0-L
19 passenger car diesel engine connected to a dynamometer, the operation of which simulated
20 city traffic driving conditions. This condensate was separated by liquid-liquid distribution
21 into hydrophilic and hydrophobic fractions representing 25 and 75% of the total condensate,
22 respectively. The hydrophilic, hydrophobic, or reconstituted hydrophobic fractions were
23 surgically implanted into the lungs of the rats. Untreated controls, vehicle
24 (beeswax/trioctanoin) controls, and positive (B[a]P) controls were also included in the
25 protocol (Table 7-2). Results of the various treatments are presented in Table 7-2. Fraction
26 lib (made up of PAHs with four to seven rings), which accounted for only 0.8% of the total
27 weight of diesel exhaust condensate, produced the highest incidence of lung carcinomas
28 following implantation into the rat lungs. A carcinoma incidence of 17.1% was observed
29 following implantation of 0.21 mg lib/rat, whereas the nitro-PAH fraction (lid) at
30 0.18 mg/rat accounted for only a 2.8% carcinoma incidence. Hydrophilic fractions of the
31 diesel exhaust paniculate extracts, vehicle (beeswax/trioctanoin) controls, and untreated
December 1994 7_25 DRAFT-DO NOT QUOTE OR CITE
-------
I
CD
o
H
O
TABLE 7-2. TUMOR INCIDENCE
WITH FRACTIONS FROM DIESEL
AND SURVIVAL TIME OF RATS TREATED
EXHAUST CONDENSATE (35 RATS/GROUP)
Material Portion by Weight (%)
Hydrophilic fraction® (25)
Hydrophobic fraction (II) (75)
Nonaromatics +
PAC 2 + 3 rings (Ha) (72)
PAH 4 to 7 rings (lib) (0.8)
Polar PAC (He) (1.1)
Nitro-PAH (lid) (0.7)
Reconstituted hydrophobics
(la), b, c, d) (74.5)
Control, unrelated
Control (beeswax/trioctanoin)
Benzo[a]pyrene
Dose (mg)
6.70
20.00
19.22
0.21
0.29
0.19
19.91
0.3
0.1
0.03
Median Survival
Time in Weeks
(range)
97(24-139)
99(50-139)
103(25-140)
102(50-140)
97(44-138)
106(32-135)
93(46-136)
110(23-138)
103(51-136)
69(41-135)
98(22-134)
97(32-135)
Number of
Carcinomas3
0
5
0
6
0
1
7
0
0
27
11
3
Number of
Adenomasb
1
0
1
0
0
0
1
0
1
0
0
0
Carcinoma
Incidence (%)
0
14.2
0
17.1
0
2.8
20.0
0
0
77.1
31.4
8.6
T^ "Squamous cell carcinoma.
5 bBronchiolar/alveolar adenoma.
Source: Adapted from Grimmer et al. (1987).
o
^
o
-------
1 controls failed to exhibit carcinoma formation. Administration of all hydrophobic fractions
2 (Ila-d) produced a carcinoma incidence (20%), similar to the summed incidence of fraction
3 lib (17.1%) and lid (2.8%). The B[a]P positive controls (0.03, 0.1, and 0.3 mg/rat) yielded
4 a carcinoma incidence of 8.6, 31.4, and 77.1%, respectively. The study showed that the
5 tumorigenic agents were primarily 4- to 7-ring PAHs and, to a lesser extent, nitroaromatics.
6 However, these studies demonstrated that simultaneous administration of various PAH
7 compounds resulted in a varying of the tumorigenic effect, thereby implying that the
8 tumorigenic potency of PAH mixtures may not depend on any one individual PAH. This
9 study did not provide any information regarding the bioavailability of the particle-associated
10 PAHs that might be responsible for carcinogenicity.
11 Kawabata et al. (1986) compared the effects of activated carbon and diesel exhaust on
12 lung tumor formation. One group of 59 F344 rats was intratracheally instilled with diesel
13 particles (1 mg/week for 10 weeks). A second group of 31 rats was instilled with the same
14 dosing regime of activated carbon. Twenty-seven rats received only the solvent (buffered
15 saline with 0.05% Tween 80), whereas 53 rats were uninjected. Rats dying after 18 mo
16 were autopsied. All animals surviving 30 mo or more postinstillation were sacrificed and
17 evaluated for histopathology. Among 42 animals exposed to diesel paniculate matter
18 surviving 18 mo or more, tumors were reported in 31, including 20 malignancies. In the
19 subgroup surviving for 30 mo, tumors were detected in 19 of 20 animals, including
20 10 malignancies. Among the rats exposed to activated carbon, the incidence of lung tumors
21 equalled 11 of 23 autopsied, with 7 cases of malignancy. Data for those dying between
22 18 and 30 mo and those sacrificed at 30 mo were not reported separately. Statistical analysis
23 indicated that activated carbon induced a significant increase in lung tumor incidence
24 compared with no tumors in 50 uninjected controls and one tumor in 23 solvent-injected
25 controls. The tumor incidence increase was significant in the diesel exposed group, and was
26 significantly greater than the increase in the carbon exposed group. This study provides
27 evidence for the carcinogenicity of diesel particulate matter. It also shows, as did Heinrich
28 (1990) for inhalation exposure, that particles lacking in organic constituents can also induce
29 tumors.
30
31
December 1994 7_27 DRAFT-DO NOT QUOTE OR CITE
-------
1 7.3.2 Syrian Hamster Studies
2 Kunitake et al. (1986) and Ishinishi et al. (1988a) conducted a study in which total
3 doses of 1.5, 7.5, or 15 mg of a dichloromethane extract of diesel exhaust was instilled
4 intratracheally over 15 weeks into male Syrian hamsters that were then held for their
5 lifetimes. The tumor incidence of 2.3% (1/44), 0% (0/56), and 1.7% (1/59) for the high-,
6 medium-, and low-dose groups, respectively, did not differ significantly from the 1.7%
7 (1/56) reported for controls. Addition of 7.5 mg of B[fl]P to an exhaust extract dose of
8 1.5 mg resulted in a total tumor incidence of 91.2% and malignant tumor incidence of 88%.
9 Benzo[a]pyrene (7.5 mg over 15 weeks) alone produced a tumor incidence rate of 88.2%
10 (85% of these being malignant), which was not significantly different from the exhaust
11 extract + B[a]P group. Intratracheal administration of 0.03 /Ltg B[a]P, the equivalent content
12 in 15 mg of exhaust extract, failed to cause a significant increase in tumors in rats. This
13 study demonstrated a lack of detectable interaction between exhaust extract and B[a]P, the
14 failure of exhaust extract to induce carcinogenesis, and the propensity for respiratory tract
15 carcinogenesis following intratracheal instillation of high doses of B[a]P. For studies using
16 the exhaust extract, some concern must be registered regarding the known differences in
17 chemical composition between exhaust extract and whole diesel exhaust. As with all
18 intratracheal instillation protocols, DE lacks the complement of volatile chemicals found in
19 whole diesel exhaust.
20 The effects on hamsters of intratracheally instilled whole diesel exhaust suspension,
21 diesel particles with Fe2O3, or diesel particle extract with Fe2O3 as the carrier were studied
22 by Shefner et al. (1982). The diesel exhaust component in each of the treatments was
23 administered at concentrations of 1.25, 2.5, or 5.0 mg/week for 15 weeks to groups of
24 50 male Syrian Golden hamsters. The total volume instilled was 3.0 mL (0.2 mL/week for
25 15 weeks). The diesel particles and the dichloromethane extracts were suspended in
26 physiological saline with gelatin (0.5% w/v), gum arable (0.5% w/v), and propylene glycol
27 (10% by volume). The Fe2O3 concentration, when used, was 1.25 mg/0.2 mL of
28 suspension. Controls received vehicle and, where appropriate, carrier particles (Fe203)
29 without the exhaust component. Two replicates of the experiments were performed.
30 Adenomatous hyperplasia was reported to be most severe in those animals treated with the
31 diesel exhaust particles or the diesel exhaust particles plus Fe2O3 particles and least severe in
December 1994 7-28 DRAFT-DO NOT QUOTE OR CITE
-------
1 those animals receiving the diesel particle extract plus Fe203. Of the two lung adenomas
2 detected microscopically, one was in a high-dose diesel-particle-treated animal and the other
3 was in a high-dose diesel extract suspension-treated animal. Although lung damage was
4 increased by instillation of diesel particles, there was no evidence of tumorigenicity.
5
6
7 7.4 SUBCUTANEOUS AND INTRAPERITONEAL INJECTION
8 STUDIES
9 7.4.1 Mouse Studies
10 In addition to inhalation studies, Orthoefer et al. (1981) also tested the effects of
11 ip injections of diesel exhaust particulate matter on male Strong A mice. Three groups of
12 30 mice were injected with 0.1 mL of a suspension (particles in distilled water) containing
13 47, 117, or 235 jug of diesel exhaust particles collected from Fluoropore filters in the
14 inhalation exposure chambers. The exposure system and exposure atmosphere are described
15 in Appendix A. Vehicle controls received injections of particle suspension made up of
16 particulate matter from control exposure filters, positive controls received 20 mg of urethan,
17 and negative controls received no injections. Injections were made three times weekly for
18 8 weeks, resulting in a total diesel particle dose of 1.1, 2.8, and 5.6 mg for the low-,
19 medium-, and high-dose groups, respectively, and 20 mg of urethan for the positive control
20 group. These animals were sacrificed after 26 weeks and examined for lung tumors. For
21 the low-, medium-, and high-diesel exhaust particle dose groups, the tumor incidence was
22 2/30, 10/30, and 8/30, respectively. The incidence among urethan-treated animals (positive
23 controls) was 100% (29/29), with multiple tumors per animal. The tumor incidence for the
24 diesel exhaust-treated animals did not differ significantly from that of vehicle controls (8/30)
25 or negative controls (7/28). The numbers of tumors per mouse was also unaffected by
26 treatment.
27 In further studies conducted by Orthoefer et al. (1981), an attempt was made to
28 compare the potency of diesel exhaust with those of other environmental pollutants. Male
29 and female Strain A mice were injected ip three times weekly for 8 weeks with diesel
30 emission particles, particle extracts, or various environmental mixtures of known
31 carcinogenicity, including cigarette smoke condensate, coke oven emissions, and roofing tar
December 1994 7-29 DRAFT-DO NOT QUOTE OR CITE
-------
1 emissions. Injection of urethan or dimethylsulfoxide (DMSO) served as positive or vehicle
2 controls, respectively. In addition to paniculate matter from the Nissan diesel previously
3 described (Section 7.2.2), an 8-cyUnder Oldsmobile engine operated at the equivalent of
4 40 mph was also used for comparison of emission effects from different makes and models
5 of diesel engine. The mice were sacrificed at 9 mo of age and their lungs examined for
6 histopathological changes. The only significant findings, other than for positive controls,
7 were small increases in numbers of lung adenomas per mouse in male mice injected with
8 Nissan diesel engine exhaust extract and in female mice injected with coke oven extract.
9 Furthermore, the increase in the extract-treated mice was significant only in comparison with
10 uninjected controls, (not injected ones) and did not occur when the experiment was repeated.
11 Despite the use of a strain of mouse known to be sensitive to tumor induction, the overall
12 findings of this study were negative. The authors provided several possible explanations for
13 these findings, the most likely of which were (1) the carcinogens that were present were very
14 weak or (2) the concentrations of the active components reaching the lungs was insufficient
15 to produce positive results.
16 Kunitake et al. (1986) conducted studies using an extract of exhaust obtained from a
17 HD, 1983, MMC M-6D22P, 11-L V-6 engine. Five sc injections of exhaust extract
18 (500 mg/kg per injection) resulted in a significant (p < 0.01) increase in subcutaneous
19 tumors for female C57B1 mice (5/22 [22.7%] versus 0/38 among controls). Five sc doses of
20 exhaust extract of 10, 25, 30, 100, or 200 mg/kg failed to produce a significant increase in
21 tumor incidence. One of 12 female ICR mice (8.3%) and 4 of 12 male ICR mice (33.3%)
22 developed malignant lymphomas following neonatal sc administration of 10 mg of exhaust
23 extract per mouse. The increase in malignant lymphoma incidence for the male mice was
24 statistically significant at (p < 0.05) compared with an incidence of 2/14 (14.3%) among
25 controls. Treatment of either sex with 2.5 or 5 mg of exhaust extract per mouse did not
26 result in statistically significant increases in tumor incidence.
27 Additional studies using paniculate matter extract from LD (1.8-L, 4-cylinder) as well
28 as HD engines with female ICR and nude mice (BALB/c/cA/JCL-nu) were also reported
29 (Kunitake et al., 1988). Groups of 30 ICR and nude mice each were given a single
30 sc injection of 10 mg HD extract, 10 mg HD + 50 /xg 12-O-tetradecanoylphorbol 13-acetate
31 (TPA), 10 mg LD extract + 50 jig TPA, or 50 /*g TPA. No malignant tumors or
December 1994 7.30 DRAFT-DO NOT QUOTE OR CITE
-------
1 papillomas were observed. One papillomatous lesion was observed in an ICR mouse
2 receiving LD extract + TPA, and acanthosis was observed in one nude mouse receiving only
3 TPA.
4 In what appears to be an extension of the Kunitake et al. (1986) sc injection studies,
5 Takemoto et al. (1988) presented additional data for subcutaneously administered extract of
6 diesel exhaust from HD and LD diesel engines. In this report, the extracts were
7 administered to 5-week-old and neonatal (<24-h-old) C57B1 mice of both sexes. Exhaust
8 particle extract from HD or LD engines was administered weekly to the 5-week-old mice for
9 5 weeks at doses of 10, 25, 50, 100, 200, or 500 mg/kg, with group sizes ranging from
10 15 to 54 animals. After 20 weeks, comparison with a control group indicated a significant
11 increase in the incidence of subcutaneous tumors for the 500-mg/kg HD group (5 of 22 mice
12 [22.7%], p < 0.01), the 100 mg/kg LD group (6 of 32 [18.8%], p < 0.01), and the
13 500 mg/kg LD group (7 of 32 [21.9%], p < 0.01) in the adult mouse experiments. The
14 tumors were characterized as malignant fibrous histiocytomas. No tumors were observed in
15 other organs. The neonates were given single doses of 2.5, 5, or 10 mg-extract
16 subcutaneously within 24 h of birth. There was a significantly higher incidence of malignant
17 lymphomas in males receiving 10 mg of HD extract and of lung tumors for males given
18 2.5 mg HD extract and for males given 5 mg and females given 10 mg LD extract.
19 A dose-related trend that was not significant was observed for the incidences of liver tumors
20 for both the HD- and LD-treated neonatal mice. The incidence of mammary tumors in
21 female mice and multiple-organ tumors in male mice was also greater for some
22 extract-treated mice but was not dose related. The report concluded that LD paniculate
23 extract showed greater carcinogenicity than did HD paniculate extract.
24
25
26 7.5 DERMAL STUDIES
27 7.5.1 Mouse Studies
28 In one of the earliest studies of diesel emissions, the effects of dermal application of
29 extract from diesel exhaust particles was examined by Kotin et al. (1955). Acetone extracts
30 were prepared from the exhaust soot of a diesel engine (type and size not provided) operated
31 at warm-up mode and under load. These extracts were applied dermally, three times weekly
December 1994 7.3! DRAFT-DO NOT QUOTE OR CITE
-------
1 to male and female C57BL and Strain A mice. Results of these experiments are summarized
2 in Table 7-3. In the initial experiments using 52 (12 male, 40 female) C57BL mice treated
3 with exhaust extract from an engine operated in a warm-up mode, two papillomas were
4 detected after 13 mo. Four tumors in 8 surviving of 50 exposed male Strain A mice treated
5 with exhaust extract from an engine operated under full load were detected 16 mo after the
6 start of treatment. For female Strain A mice treated with extract from an engine operated
7 under full load, 17 tumors were detected in 20 of 25 mice surviving longer than 13 mo.
8 This provided a significantly increased tumor incidence of 85%. Carcinomas as well as
9 papillomas were seen, but the numbers were not reported.
10 Depass et al. (1982) examined the potential of diesel exhaust particles and
11 dichloromethane extracts of diesel exhaust particles to act as complete carcinogens,
12 carcinogen initiators, or carcinogen promoters. In skin-painting studies, the exhaust material
13 was obtained from an Oldsmobile 5.7-L diesel engine operated under constant load at
14 65 km/h. The exhaust particles were collected at a temperature of 100 °C. Groups of
15 40 C3H/HeJ mice were used because of their low spontaneous tumor incidence. For the
16 complete carcinogenesis experiments, diesel exhaust particles were applied as a 5 or 10%
17 suspension in acetone. Dichloromethane extract was applied as 5, 10, 25, or 50%
18 suspensions. Negative controls received acetone, and positive controls received 0.2% B[a]P.
19 For tumor-promotion experiments, a single application of 1.5% B[a]P was followed by
20 repeated applications of 10% diesel particle suspension, 50% diesel particle extract, acetone
21 only (vehicle control), 0.0001% phorbol 12-myristate 13-acetate (PMA) as a positive
22 promoter control, or no treatment (negative control). For the tumor-initiation studies, a
23 single initiating dose of 10% diesel particle suspension, 50% diesel particle extract, acetone,
24 or PMA was followed by repeated applications of 0.0001 % PMA. Following 8 mo of
25 treatment, the PMA dose in the initiation and promotion studies was increased to 0.01%.
26 Animals were treated three times per week in the complete carcinogenesis and initiation
27 experiments and five times per week in promotion experiments. All test compounds were
28 applied to a shaved area on the back of the mouse.
29 In the complete carcinogenesis experiments, one mouse receiving the high-dose (50%)
30 suspension of extract developed a squamous cell carcinoma after 714 days of treatment.
31 Tumor incidence in the E[a]P group was 100% and no tumors were observed in any of the
December 1994 7.32 DRAFT-DO NOT QUOTE OR CITE
-------
B
a
H
W
o
H
a
H
u
3 H
SS
u
u
I
o
2
o
ti
o a
CH C
Q .§
&
C/l
3 I
o 8
>
P!
UH
i_i
o
2
l?l
le Material
00
ts
(S "H
en
00
CO
m
"c3 73
•C 'C
4> 4i
e« g< cj
d 3 a
s
p w w
rt (3 rt «^j rt 3
g ^ -ga -ga
•rt W) -p W) -p M
•" OJJ
si
&1
tj_i T-l
g .5 g .5
is 43 •» S
W "o W "o
:^ .a
r~ «»
U U S
UH
on
IT)
rs
ON
•a
.a
i
I
December 1994
7-33
DRAFT-DO NOT QUOTE OR CITE
-------
1 other groups. For the promotion studies, squamous cell carcinomas with pulmonary
2 metastases were identified in one mouse of the 50% diesel exhaust particle extract group, and
3 one in the 25% extract group. Another mouse in the 25% extract group developed a grossly
4 diagnosed papilloma. Nineteen positive control mice had tumors (11 papillomas,
5 8 carcinomas). No tumors were observed for any of the other treatment groups. For the
6 initiation studies, three tumors (two papillomas and one carcinoma) were identified in the
7 group receiving diesel particle suspension and three tumors (two papillomas and one
8 fibrosarcoma) were found in the diesel exhaust particle extract group. These findings were
9 reported to be statistically insignificant using the Breslow and Mantel-Cox tests.
10 The data from this study indicated that diesel exhaust particles and dichloromethane
11 extracts of these particles are not effective with regard to tumor promotion or initiation.
12 Although these findings were not consistent with those of Kotin et al. (1955) (Table 7-3), the
13 occurrence of a single carcinoma in a strain known to have an extremely low spontaneous
14 tumor incidence may be of importance. Furthermore, a comparison between studies
15 employing different strains of mice with varying spontaneous tumor incidences may result in
16 erroneous assumptions.
17 Nesnow et al. (1982) studied the formation of dermal papillomas and carcinomas
18 following dermal application of dichloromethane extracts from coke oven emissions, roofing
19 tar, diesel engine exhaust, and gasoline engine exhaust. Diesel exhaust from five different
20 engines including a preproduction Nissan 220C, a 5.7-L Oldsmobile, a prototype VW Turbo
21 Rabbit, a Mercedes 300D, and a HD Caterpillar 3304 were used for various phases of the
22 study. Male and female Sencar mice (40 per group) were used for tumor-initiation,
23 tumor-promotion, and complete carcinogenesis studies. For the tumor-initiation experiments,
24 the diesel exhaust extracts were topically applied in single doses of 100, 500, 1,000 or
25 2,000 /ig/mouse. The high dose (10,000 /ig/mouse) was applied in five daily doses of
26 2,000 /ig. One week later, 2 /xg of the tumor promoter tetradecanoylphorbol acetate (TPA)
27 was applied topically twice weekly. The tumor-promotion experiments used mice treated
28 with 50.5 /*g of B[a]P followed by weekly (twice weekly for high dose) topical applications
29 (at the aforementioned doses) of the extracts. For the complete carcinogenesis experiments,
30 the test extracts were applied weekly (twice weekly for the high doses) for 50 to 52 weeks.
December 1994 7.34 DRAFT-DO NOT QUOTE OR CITE
-------
1 Only extracts from the Nissan, Oldsmobile, and Caterpillar engines were used in the
2 complete carcinogenesis experiments.
3 In the tumor-initiation studies, both B[a]P alone and the Nissan engine exhaust extract
4 followed by TPA treatment produced a significant increase in tumor (dermal papillomas)
5 incidence at 7 to 8 weeks postapplication. By 15 weeks, the tumor incidence was greater
6 than 90% for both groups. No significant carcinoma formation was noted for mice in the
7 tumor-initiation experiments following exposure to the exhaust extracts of the other diesel
8 engines, although the Oldsmobile engine exhaust extract at 2.0 mg/mouse did
9 produce a 40% papilloma incidence in male mice at 6 mo. This effect was, however, not
10 dose dependent.
11 Benzo[fl]pyrene (50.5 /xg/week), coke oven extract (at 1.0, 2.0, or 4.0 mg/week), and
12 the highest dose of roofing tar extract (4.0 mg/week) all tested positive for complete
13 carcinogenesis activity. Exhaust extracts from only the Nissan, Oldsmobile, and Caterpillar
14 engines were tested for complete carcinogenic potential, and all three proved to be negative
15 using the Sencar mouse assay.
16 The results of the dermal application experiments by Nesnow et al. (1982) are
17 presented in Table 7-4. The tumor initiation-promotion assay was considered positive if a
18 dose-dependent response was obtained and if at least two doses provided a
19 papilloma-per-mouse value that was three times or greater than that of the background value.
20 Based on these criteria, only emissions from the Nissan were considered positive. Tumor
21 initiation and complete carcinogenesis assays required that at least one dose produce a tumor
22 incidence of at least 20%. None of the diesel exhaust samples yielded positive results based
23 on this criterion.
24 Kunitake et al. (1986, 1988) evaluated the effects of a dichloromethane extract of diesel
25 exhaust particulate matter obtained from a 1983 MMC M-6D22P, 11-L V-6 engine.
26 An acetone solution was applied in 10 doses every other day, followed by promotion with
27 2.5 /^g of TPA, three times weekly for 25 weeks. Exposure groups received a total dose of
28 0.5, 5, 15, or 45 mg of extract. Papillomas were reported in 2 of 50 animals examined in
29 the 45-mg exposure group and 1/48 in the 15-mg group compared with 0/50 among controls.
30 Differences, however, were not statistically significant.
December 1994 7.35 DRAFT-DO NOT QUOTE OR CITE
-------
December 19!
E
2
ON
O
3
O
0
O
1
w
0
n
TABLE 7-4. DERMAL
Sample
Benzo[a]pyrene
Topside coke oven
Coke oven main
Roofing tar
Nissan
Oldsmobile
VW Rabbit
Mercedes
Caterpillar
Residential furnace
Mustang
aScored at 6 mo.
bCumulative score at 1 year.
cMale/female.
dND = Not determined.
el = Incomplete.
Source: Nesnow et al. (1982).
TUMORIGENIC AND CARCINOGENIC EFFECTS OF VARIOUS EMISSION EXTRACTS
Tumor Initiation Complete Carcinogenesis Tumor Promotion
Papillomas3 Carcinomas'5 Carcinomas'5 Papilomas3
+/+c +/+ +/+ +/+
+/+ -/+ NDd ND
+/+ +/+ +/+ +/+
+/+ +/+ +/+ +/+
+/+ +/+ -/- . ND
+/+ -/- -/- ND
+/+ -/- r ND
+/- -/- ND ND
-/- -/- -/- ND
-/- -/- ND ND
+/+ -/+ ND ND
-------
1 7.6 SUMMARY AND CONCLUSIONS OF ANIMAL
2 CARCINOGENICITY STUDIES
3 As early as 1955, Kotin et al. (1955) provided evidence for tumorigenicity and
4 carcinogenicity of acetone extracts of diesel exhaust following dermal application and also
5 provided data suggesting a difference in this potential depending on engine operating mode.
6 Until the early 1980s, no chronic studies assessing inhalation of diesel exhaust, the relevant
7 mode for human exposure, had been reported. Since then, inhalation studies have been
8 emphasized. Studies employing rats and an experimental protocol including long-term
9 exposure at high exposure concentrations (up to 8 mg/m3), resulting in large lung particle
10 loads and a postexposure observation period, were generally positive in demonstrating diesel
11 exhaust-induced increases in tumorigenicity. The highest incidences of tumors were reported
12 by Brightwell et al. (1986). Among female rats exposed for 24 mo and held for their
13 lifetimes, tumors were detected in 24/25 animals. This study points out the probable
14 cumulative effects of high exposure concentration (6.6 mg/m3), lengthy daily exposures
15 (16 h/day), exposure in the dark resulting in a probable increase in ventilation and thereby
16 particle intake, and maintenance of the animals for their lifetimes. In two other major
17 studies, Heinrich et al. (1986a) and Mauderly et al. (1987), significant but lower lung tumor
18 incidences were reported at the high-dose levels, 15.8 and 12.8%, respectively. Although
19 exposure concentrations differed, 7 mg/m3 for Mauderly et al. versus 4 mg/m3 for Heinrich
20 et al., the longer daily exposure periods in the Heinrich et al. study, 19 h versus 7 h, would
21 probably result in only slightly differing intakes. Ishinishi et al. (1988a,b) reported a
22 6.5% incidence of lung tumors in rats exposed to a concentration of 4 mg/m3 paniculate
23 matter from a HD engine. In this study, although the concentration was relatively low,
24 duration and length of daily exposure was long (16 h/day for 30 mo). Iwai et al. (1986)
25 reported an increased lung tumor incidence (4/14) in Fischer rats exposed 8 h/day,
26 7 days/week for 24 mo to a particle concentration of 4.9 mg/m3. Four of five held in clean
27 air an additional 3 to 6 mo, however, also developed tumors pointing out again the
28 importance of a long study duration. Iwai et al. (1986) reported the only diesel exhaust
29 inhalation-induced tumor increase at a nonrespiratory site (splenic lymphoma).
30 Low exposure concentrations and/or short exposure durations were generally used in
31 the negative studies (Karagianes et al., 1981; Lewis et al., 1986; White et al., 1983;
December 1994 7.37 DRAFT-DO NOT QUOTE OR CITE
-------
1 Takemoto et al., 1986). The lowest particle concentrations resulting in significant positive
2 effects in rats were in the range of 2 to 3 mg/m3.
3 Inhalation of diesel exhaust induced significant increases in lung tumors in female
4 NMRI mice (Heinrich et al., 1986a; Stober, 1986) and in female Sencar mice (Pepelko and
5 Peirano, 1983). An apparent increase was also seen in female C57BL mice (Takemoto
6 et al., 1986). In a series of short-term inhalation studies using Strain A, mice no increases
7 in lung tumor rates were detected (Orthoefer et al., 1981; Kaplan et al., 1982, 1983; White
8 et al., 1983). The only study in which lung tumor incidences were increased in animals
9 exposed to filtered exhaust was reported by Heinrich et al. (1986a) and Stober (1986) using
10 NMRI mice.
11 Attempts to induce significant increases in lung tumors in Syrian hamsters were
12 unsuccessful after inhalation (Heinrich et al., 1982; Heinrich et al., 1986a; Heinrich et al.,
13 1989b; Brightwell et al., 1986) or intratracheal instillation (Kunitake et al., 1986; Ishinishi
14 et al., 1988a). Neither cats (Pepelko and Peirano, 1983 [see Chapter 4]), nor monkeys
15 (Lewis et al., 1986) developed tumors following at 2-year exposure to diesel exhaust. The
16 duration of these exposures, however, may well have been inadequate in these two
17 longer-lived species. Exposure levels were also below the MTD in the monkey studies, and
18 borderline for detection of lung tumor increases in rats.
19 Kawabata et al. (1986) demonstrated the induction of lung tumors in Fischer 344 rats
20 following intratracheal instillation of diesel paniculate matter. Grimmer et al. (1987)
21 showed, not only that an extract of diesel particles was carcinogenic when instilled in the
22 lungs of rats but also that most of the carcinogenicity resided in the portion containing PAHs
23 with four to seven rings.
24 Alternative exposure routes including dermal exposure and sc injection in mice
25 provided additional evidence for tumorigenic effects of diesel exhaust. Particle extracts
26 applied dermally to mice have been shown to induce significant skin tumor increases in two
27 studies (Kotin et al., 1955; Nesnow et al., 1982). Kunitake et al. (1986) also reported a
28 marginally significant increase in skin papillomas in ICR mice treated with an organic extract
29 from an HD diesel engine. Negative results were reported by Depass et al. (1982) for
30 skin-painting studies using mice and acetone extracts of diesel exhaust particle suspensions.
31 However, in this study the exhaust particles were collected at temperatures of 100 °C,
December 1994 7.33 DRAFT-DO NOT QUOTE OR CITE
-------
1 a temperature that would minimize the condensation of vapor-phase organics and, therefore,
2 reduce the availability of potentially carcinogenic compounds that might normally be present
3 on diesel exhaust particles. A significant increase in the incidence of sarcomas in female
4 C57B1 mice was reported by Kunitake et al. (1986) following sc administration of LD diesel
5 exhaust particle extract at doses of 500 mg/kg. Takemoto et al. (1988) provided additional
6 data for this study and reported an increased tumor incidence in the mice following injection
7 of LD engine exhaust extract at doses of 100 and 500 mg/kg. Results of ip injection of
8 diesel exhaust paniculate matter or particle extracts in strain A mice were generally negative
9 (Orthoefer et al., 1981; Pepelko and Peirano, 1983), suggesting that the strain A mouse may
10 not be a good model for testing of diesel emissions.
11 Experiments using tumor initiators such as DEN, B[0]P, DPN, or DBA (Brightwell
12 et al., 1986; Heinrich et al., 1986a; Takemoto et al., 1986) did not provide conclusive
13 results regarding the tumor promoting potential of either filtered or whole diesel exhaust.
14 A report by Heinrich et al. (1982), however, did indicate that filtered exhaust may promote
15 the tumor-initiating effects of DEN in hamsters.
16 Several reports (Wong et al., 1986; Bond et al., 1990) affirm observations of the
17 potential carcinogenicity of diesel exhaust by providing evidence for DNA damage in rats.
18 These findings are discussed in more detail in Chapter 9. Evidence for the mutagenicity of
19 organic agents present in diesel engine emissions is also provided in Chapter 8.
20 It appears reasonably certain that with adequate exposures, inhalation of diesel exhaust
21 will induce lung cancer in rats and in at least some strains of mice. The relationship between
22 exposure levels and response, however, is less clearcut. Although significant increases in
23 lung tumors were not reported at concentrations less than about 2 mg/m3, the response at
24 higher concentrations varies considerably. A significant percentage of this variation can
25 probably be attributed to the exposure regime. A better method than concentration alone for
26 assessing exposure-response relationships could be achieved by comparing cumulative
27 exposure (concentration x daily exposure duration X days of exposure). Only those studies
28 conducted for a sufficient length of time (>24 mo) for expression of carcinogenic responses
29 have been included in this analysis. Examination of the rat data, shown in Table 7-5 and
30 plotted in Figure 7-1 reveals that most studies indicate a trend of increasing tumor incidence
31 at exposures exceeding 1 x 104 mg-h/m3.
December 1994 7.39 DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 7-5. CUMULATIVE (CONCENTRATION x TIME) EXPOSURE DATA
FOR RATS EXPOSED TO WHOLE DIESEL EXHAUST
1 •
n>
1-1
53 Study
£ Mauderly et al. (1987)
Heinrich et al. (1986b)
Ishinishi et al. (1988b)
(Light-duty engine)
T4
o
(Heavy-duty engine)
O
£•
H
1
0
o
2 Brightwell et al. (1989)
O
H
tO
a
o
H
w
0
90
0
Exposure
Rate/Duration
(h/week, mo)
35, 30
35,30
35, 30
35,30
95,35
95,35
96,30
96,30
96,30
96,30
96, 30
96,30
96, 30
96, 30
96,30
96, 30
80,24
80,24
80,24
80, 24
Total Exposure
Time
00
4,200
4,200
4,200
4,200
13,300
13,300
11,520
11,520
11,520
11,520
11,520
11,520
11,520
11,520
11,520
11,520
7,680
7,680
7,680
7,680
Particle
Concentration
(mg/m3)
0
0.35
3.5
7.1
0
4.24
0
0.1
0.4
1.1
2.3
0
0.5
1.0
1.8
3.7
0
0.7
2.2
6.6
Cumulative Exposure
(mg -h/m3)
Per Week
0
12.25
122.5
248.5
0
402.8
0
9.6
38.4
105.6
220.8
0
48.0
96.0
172.8
355.2
0
56.0
176.0
528.0
Total
0
1,470
14,700
29,820
0
56,392
0
1,152
4,608
12,672
26,496
0
5,760
11,520
20,736
42,624
0
5,376
16,896
50,688
Tumor Incidence
(%)a
0.9
1.3
3.6
12.8
0
17.8
3.3
2.4
0.8
4.1
2.4
0.8
0.8
0
3.3
6.5
1.2
0.7
9.7
38.5
-------
TABLE 7-5 (cont'd). CUMULATIVE (CONCENTRATION x TIME) EXPOSURE DATA
FOR RATS EXPOSED TO WHOLE DIESEL EXHAUST
-J
O
%
3
6
o
I
0
o
a
i
0
Exposure
Rate/Duration
Study (h/week, mo)
Kaplan et al. (1983) 140, 15
140, 15
140, 15
140, 15
Iwai et al. (1986) 56, 24
56,24
Takemoto et al. (1986) 16, 18-24
16, 18-24
Karagianes et al. (1981) 30, 20
30, 20
aCombined data for males and females.
Total Exposure
Time
CO
8,400
8,400
8,400
8,400
5,376
5,376
1,152-1,536
1,152-1,536
2,400
2,400
Particle
Concentration
(mg/m3)
0
0.25
0.75
1.5
0
4.9
0
2-4
0
8.3
Cumulative Exposure
(mg -h/m3)
Per Week Total
0
35
105
210
0
274.4
0
32-64
0
249
0
2,100
6,300
12,600
0
26,342
0
3,456-4,608
0
19,920
Tumor Incidence
0
3.3
10.0
3.3
0
36.8
0
0
0
16.6
-------
40
30
O Mauderlyetal. (1987)
• Heinrichetal. (1986b)
V Ishinishi et al. (19885) (LD)
^ Ishinishi et al. (1988b) (HD)
D Brightwell eta). (1989)
A Iwaietal. (1986)
§
-------
1 models (Appendix A) and the qualitative/quantitative evaluations of Chapter 11 attempt this
2 relative to human exposure.
3 To evaluate accurately the carcinogenic risk to humans from diesel engine emissions it
4 is important to ascertain the fraction or fractions of exhaust responsible for induction of lung
5 tumors. Several of the previously discussed studies indicated that only whole (unfiltered)
6 diesel exhaust is tumorigenic or carcinogenic and that these properties are eliminated or
7 greatly minimized in filtered diesel exhaust exposure. In one study (Stober, 1986), however,
8 a significant increase in lung tumors was seen in mice exposed to filtered exhaust. Heinrich
9 et al. (1982) also provided some evidence suggesting that the gaseous fraction promoted the
10 tumorigenic effects of DEN. Nevertheless, because of the lack of positive data in rats and
11 the limited positive data in mice, the tumorigenicity of the gaseous fraction must be
12 considered to be unresolved.
13 The relative contribution of the carbon core of the diesel particles versus organics
14 adsorbed to the surface of the particles to cancer induction is still somewhat uncertain. The
15 primary evidence for the importance of the adsorbed organics is the presence of known
16 carcinogens among these chemicals. These include polycyclic aromatics as well as
17 nitroaromatics as described in Chapters 2 and 3. Organic extracts of particles have also been
18 shown to induce tumors in a variety of injection, intratracheal instillation and skin painting
19 studies, and Grimmer et al. (1987) has, in fact, shown that the great majority of the
20 carcinogenic potential following intratracheal instillation resided in the fraction containing
21 four- to seven-ring PAHs.
22 Evidence for the importance of the carbon core is provided by studies of Kawabata
23 et al. (1986), that showed induction of lung tumors following intratracheal instillation of
24 CB that contained no more than traces of organics and studies of Heinrich (1990) that
25 indicated that exposure via inhalation to CB(Printex 90) particles induced lung tumors at
26 concentrations similar to those effective in diesel studies. Other particles of low solubility
27 such as titanium dioxide (Lee et al., 1986) have also been shown to induce lung tumors,
28 although at much higher concentrations than necessary for carbon particles or diesel exhaust.
29 Pyrolyzed pitch, on the other hand, essentially lacking a carbon core but having PAH
30 concentrations at least three orders of magnitude greater than diesel exhaust, was no more
31 effective in tumor induction than was diesel exhaust (Heinrich et al., 1986b). These studies
December 1994 7.43 DRAFT-DO NOT QUOTE OR CITE
-------
1 suggest that the insoluble carbon core of the particle is at least as important as the organic
2 components and possibly more so for lung tumor induction at high particle concentrations
3 (>2 mg/m3). A more detailed discussion of this issue can be found in Chapter 11.
4 In summary, based on positive inhalation exposure data in rats and mice, intratracheal
5 instillation in rats, and injection or skin painting in mice and supported by positive
6 mutagenicity studies, the evidence for carcinogenicity of diesel exhaust is considered to be
7 adequate. The contribution of the various fractions of diesel exhaust to the carcinogenic
8 response is less certain. The effects of the gaseous phase are equivocal. The presence of
9 known carcinogens adsorbed to diesel particles and the demonstrated tumorigenicity of
10 particle extracts in a variety of injection, instillation- and skin-painting studies provides
11 evidence for the involvement of the organic fraction. Studies showing that pure carbon
12 particles can also induce tumors, on the other hand, indicate that the carbon core of the
13 diesel particle is also involved in the carcinogenic process.
14 A summary of studies assessing the tumorigenic and carcinogenic effects in laboratory
15 animals following inhalation exposure to diesel exhaust is presented in Table 7-1.
December 1994 7.44 DRAFT-DO NOT QUOTE OR CITE
-------
1 REFERENCES
2 Bond, J. A.; Johnson, N. F.; Snipes, M. B.; Mauderly, J. L. (1990) DNA adduct formation in rat alveolar
3 type II cells: cells potentially at risk for inhaled diesel exhaust. Environ. Mol. Mutagen. 16: 64-69.
4
5 Brightwell, J.; Fouillet, X.; Cassano-Zoppi, A.-L.; Gatz, R.; Duchosal, F. (1986) Neoplastic and functional
6 changes in rodents after chronic inhalation of engine exhaust emissions. In: Ishinishi, N.; Koizumi, A.;
7 McClellan, R. O.; Stober, W., eds. Carcinogenic and mutagenic effects of diesel engine exhaust:
8 proceedings of the international satellite symposium on toxicological effects of emissions from diesel
9 engines; July; Tsukuba Science City, Japan. Amsterdam, Holland: Elsevier Science Publishers B. V.;
10 pp. 471-485. (Developments in toxicology and environmental science: v. 13).
11
12 Brightwell, J.; Fouillet, X.; Cassano-Zoppi, A.-L.; Bernstein, D.; Crawley, F.; Duchosal, R.; Gatz, R.;
13 Perczel, S.; Pfeifer, H. (1989) Tumors of the respiratory tract in rats and hamsters following chronic
14 inhalation of engine exhaust emissions. J. Appl. Toxicol. 9: 23-31.
15
16 Carpenter, K.; Johnson, J. H. (1980) Analysis of the physical characteristics of diesel paniculate matter using
17 transmission electron microscope techniques. SAE Trans. 88: 2743-2759.
18
19 Cuddihy, R. G.; Griffith, W. C.; McClellan, R. O. (1984) Health risks from light-duty diesel vehicles. Environ.
20 Sci. Technol. 18: 14A-21A.
21
22 Depass, L. R.; Chen, K. C.; Peterson, L. G. (1982) Dermal carcinogenesis bioassays of diesel particulates and
23 dichloromethane extract of diesel particulates in C3H mice. In: Lewtas, J., ed. Toxicological effects of
24 emissions from diesel engines: proceedings of the Environmental Protection Agency 1981 diesel
25 emissions symposium; October 1981; Raleigh, NC. New York, NY: Elsevier Biomedical; pp. 321-326.
26 (Developments in toxicology and environmental science: v. 10).
27
28 Grimmer, G.; Brune, H.; Deutsch-Wenzel, R.; Dettbarn, G.; Jacob, J.; Naujack, K.-W.; Mohr, U.; Ernst, H.
29 (1987) Contribution of polycyclic aromatic hydrocarbons and nitro-derivatives to the carcinogenic impact
30 of diesel engine exhaust condensate evaluated by implantation into the lungs of rats. Cancer Lett.
31 (Shannon, Irel.) 37: 173-180.
32
33 Heinrich, U. (1990) Results of long-term inhalation exposure of rats to carbon black "Printex 90" [letter to
34 Dr. Lester D. Grant]. Presented at: U.S. Environmental Protection Agency peer review workshop on the
35 Health Assessment Document for Diesel Emissions; July; Research Triangle Park NC.
36
37 Heinrich, U.; Peters, L.; Funcke, W.; Pott, F.; Mohr, U.; Stober, W. (1982) Investigation of toxic and
38 carcinogenic effects of diesel exhaust in long-term inhalation exposure of rodents. In: Lewtas, J., ed.
39 Toxicological effects of emissions from diesel engines: proceedings of the Environmental Protection
40 Agency diesel emissions symposium; October 1981; Raleigh, NC. New York, NY: Elsevier Biomedical;
41 pp. 225-242. (Developments in toxicology and environmental science: v. 10).
42
43 Heinrich, U.; Muhle, H.; Takenaka, S.; Ernst, H.; Fuhst, R.; Mohr, U.; Pott, F.; Stober, W. (1986a) Chronic
44 effects on the respiratory tract of hamsters, mice, and rats after long-term inhalation of high
45 concentrations of filtered and unfiltered diesel engine emissions. J. Appl. Toxicol. 6: 383-395.
46
47 Heinrich, U.; Pott, F.; Rittinghausen, S. (1986b) Comparison of chronic inhalation effects in rodents after
48 long-term exposure to either coal oven flue gas mixed with pyrolized pitch or diesel engine exhaust.
49 In: Ishinishi, N.; Koizumi, A.; McClellan, R. O.; Stober, W., eds. Carcinogenic and mutagenic effects
50 of diesel engine exhaust: proceedings of the international satellite syposium on toxicological effects of
51 emissions from diesel engines; July; Tsukuba Science City, Japan. Amsterdam, Holland: Elsevier Science
52 Publishers B. V.; pp. 441-457. (Developments in toxicology and environmental science: v. 13).
»s J
December 1994 7.45 DRAFT-DO NOT QUOTE OR CITE
-------
1 Heinrich, U.; Peters, L.; Fuhst, R.; Mohr, U. (1989a) The effect of automotive exhaust exposure on the
2 carcinogenicity of dipentylnitrosamine (DPN) in the respiratory tract of rats. Exp. Pathol. 37: 51-55.
3
4 Heinrich, U.; Mohr, U.; Fuhst, R.; Brockmeyer, C. (1989b) Investigation of a potential cotumorigenic effect of
5 the dioxides of nitrogen and sulfur, and of diesel-engine exhaust, on the respiratory tract of Syrian golden
6 hamsters. Cambridge, MA: Health Effects Institute; research report no. 26. Available from: NTIS,
7 Springfield, VA; PB90-111147.
8
9 Huisingh, J.; Bradow, R.; lungers, R.; Claxton, L.; Zweidinger, R.; Tejada, S.; Bumgarner, J.; Duffield, F.;
10 Waters, M.; Simmon, V. F.; Hare, C.; Rodriguez, C.; Snow, L. (1978) Application of bioassay to the
11 characterization of diesel particle emissions. In: Waters, M. D.; Nesnow, S.; Huisingh, J. L.; Sandhu,
12 S. S.; Claxton, L., eds. Application of short-term bioassays in the fractionation and analysis of complex
13 environmental mixtures: [proceedings of a symposium; February; Williamsburg, VA]. New York, NY:
14 . Plenum Press; pp. 383-418. (Hollaender, A.; Probstein, F.; Welch, B. L., eds. Environmental science
15 research: v. 15).
16
17 International Agency for Research on Cancer. (1989) Diesel and gasoline engine exhausts and some nitroarenes.
18 Lyon, France: World Health Organization; p. 104. (IARC monographs on the evaluation of carcinogenic
19 risks to humans: v. 46).
20
21 Ishinishi, N.; Inamasu, T.; Hisanaga, A.; Tanaka, A.; Hirata, M.; Ohyama, S. (1988a) Intratracheal instillation
22 study of diesel paniculate extracts in hamsters. In: Diesel exhaust and health risks: results of the HERP
23 studies. Tsukuba, Ibaraki, Japan: Japan Automobile Research Institute, Inc., Research Committee for
24 HERP Studies; pp. 209-216.
25
26 Ishinishi, N.; Kuwabara, N.; Takaki, Y.; Nagase, S.; Suzuki, T.; Nakajima, T.; Maejima, K.; Kato, A.;
27 Nakamura, M. (1988b) Long-term inhalation experiments on diesel exhaust. In: Diesel exhaust and health
28 risks: results of the HERP studies. Tsukuba, Ibaraki, Japan: Japan Automobile Research Institute, Inc.,
29 Research Committee for HERP Studies; pp. 11-84.
30
31 Iwai, K.; Udagawa, T.; Yamagishi, M.; Yamada, H. (1986) Long-term inhalation studies of diesel exhaust on
32 F344 SPF rats. Incidence of lung cancer and lymphoma. In: Ishinishi, N.; Koizumi, A.; McClellan,
33 R. O.; Stober, W., eds. Carcinogenic and mutagenic effects of diesel engine exhaust: proceedings of the
34 international satellite symposium on lexicological effects of emissions from diesel engines; July; Tsukuba
35 Science City, Japan. Amsterdam, Holland: Elsevier Science Publishers B. V.; pp. 349-360.
36 (Developments in toxicology and environmental science: v. 13).
37
38 Kaplan, H. L.; MacKenzie, W. F.; Springer, K. J.; Schreck, R. M.; Vostal, J. J. (1982) A subchronic study of
39 the effects of exposure of three species of rodents to diesel exhaust. In: Lewtas, J., ed. Toxicological
40 effects of emissions from diesel engines: proceedings of the Environmental Protection Agency diesel
41 emission symposium; October, 1981; Raleigh, NC. New York, NY: Elsevier Biomedical; pp. 161-182.
42 (Developments in toxicology and environmental science: v. 10).
43
44 Kaplan, H. L.; Springer, K. J.; MacKenzie, W. F. (1983) Studies of potential health effects of long-term
45 exposure to diesel exhaust emissions. San Antonio, TX: Southwest Research Institute; SwRI project
46 no. 01-0750-103.
47
48 Karagianes, M. T.; Palmer, R. F.; Busch, R. H. (1981) Effects of inhaled diesel emissions and coal dust in rats.
49 Am. Ind. Hyg. Assoc. J. 42: 382-391.
50
51
December 1994 7-46 DRAFT-DO NOT QUOTE OR CITE
-------
1 Kawabata, Y.; Iwai, K.; Udagawa, T.; Tukagoshi, K.; Higuchi, K. (1986) Effects of diesel soot on unscheduled
2 DNA synthesis of trachea! epithelium and lung tumor formation. In: Ishinishi, N.; Koizumi, A.;
3 McClellan, R. O.; St6ber, W., eds. Carcinogenic and mutagenic effects of diesel engine exhaust:
4 proceedings of the international satellite symposium on lexicological effects of emissions from diesel
5 engines; July; Tsukuba Science City, Japan. Amsterdam, The Netherlands: Elsevier Science Publishers
6 B. V.; pp. 213-222. (Developments in toxicologial and environmental science: v. 13).
7
8 Kotin, P.; Falk, H. L.; Thomas, M. (1955) Aromatic hydrocarbons: III. presence in the paniculate phase of
9 diesel-engine exhausts and the carcinogenicity of exhaust extracts. AMA Arch. Ind. Health 11:113-120.
10
11 Kunitake, E.; Shimamura, K.; Katayama, H.; Takemoto, K.; Yamamoto, A.; Hisanaga, A.; Ohyama, S.;
12 Ishinishi, N. (1986) Studies concerning carcinogenesis of diesel paniculate extracts following intratracheal
13 instillation, subcutaneous injection, or skin application. In: Ishinishi, N.; Koizumi, A.; McClellan, R. O.;
14 Stober, W., eds. Carcinogenic and mutagenic effects of diesel engine exhaust: proceedings of the
15 international satellite symposium on lexicological effects of emissions from diesel engines; July; Tsukuba
16 Science City, Japan. Amsterdam, The Netherlands: Elsevier Science Publishers B. V.; pp. 235-252.
17 (Developments in toxicology and environmental science: v. 13).
18
19 Kunitake, E.; Imase, A.; Shimamura, K.; Ishinishi, N.; Hisanaga, A.; Tanaka, A. (1988) Skin application and
20 subcutaneous injection experiments of diesel paniculate extracts using ICR mice and nude mice.
21 In: Diesel exhaust and health risks: results of the HERP studies. Tsukuba, Ibarakl, Japan: Japan
22 Automobile Research Institute, Inc., Research Committee for HERP Studies; pp. 217-225.
23
24 Lee, K. P.; Henry, N. W., Ill; Trochimowicz, H. J.; Reinhardt, C. F. (1986) Pulmonary response to impaired
25 lung clearance in rats following excessive TiO2 dust deposition. Environ. Res. 41: 144-167.
26
27 Lewis, T. R.; Green, F. H. Y.; Moorman, W. J.; Burg, J. A. R.; Lynch, D. W. (1986) A chronic inhalation
28 toxicity study of diesel engine emissions and coal dust, alone and combined. In: Ishinishi, N.;
29 Koizumi, A.; McClellan, R. 0.; Stober, W., eds. Carcinogenic and mutagenic effects of diesel engine
30 exhaust: proceedings of the international satellite symposium on lexicological effects of emissions from
31 diesel engines; July; Tsukuba Science City, Japan. Amsterdam, The Netherlands: Elsevier Science
32 Publishers B. V.; pp. 361-380. (Developments in toxicology and environmental science: v. 13).
33
34 Mauderly, J. L.; Jones, R. K.; Griffith, W. C.; Henderson, R. F.; McClellan, R. O. (1987) Diesel exhaust is a
35 pulmonary carcinogen in rats exposed chronically by inhalation. Fundam. Appl. Toxicol. 9: 208-221.
36
37 Mauderly, J. L.; Snipes, M. B.; Barr, E. B.; Bechtold, W. E.; Henderson, R. F.; Mitchell, C. E.; Nikula,
38 K. J.; Thomassen, D. G. (1991) Influence of particle-associated organic compounds on carcinogenicity of
39 diesel exhaust. Presented at: eighth Health Effects Institute annual conference; April; Colorado Springs,
40 CO. Cambridge, MA: Health Effects Institute.
41
42 Mohr, U.; Takenaka, S.; Dungworth, D. L. (1986) Morphologic effects of inhaled diesel engine exhaust on lungs
43 of rats: comparison with effects of coal oven flue gas mixed with pyrolized pitch. In: Ishinishi, N.;
44 Koizumi, A.; McClellan, R. O.; Stober, W., eds. Carcinogenic and mutagenic effects of diesel engine
45 exhaust: proceedings of the international satellite symposium on toxicological effects of emissions from
46 diesel engines; July; Tsukuba Science City, Japan. Amsterdam, Holland: Elsevier Science Publishers
47 B. V.; pp. 459-470. (Developments in toxicology and environmental science: v. 13).
48
49 Mokler, B. V.; Archibeque, F. A.; Beethe, R. L.; Kelly, C. P. J.; Lopez, J. A.; Mauderly, J. L.; Stafford,
50 D. L. (1984) Diesel exhaust exposure system for animal studies. Fundam. Appl. Toxicol. 4: 270-277.
52
December 1994 7.47 DRAFT-DO NOT QUOTE OR CITE
-------
1 Nesnow, S.; Evans, C.; Stead, A.; Creason, J.; Slaga, T. J.; Triplett, L. L. (1982) Skin carcinogenesis studies
2 of emission extracts. In: Lewtas, J., ed. Toxicological effects of emissions from diesel engines:
3 proceedings of the Environmental Protection Agency diesel emissions symposium; October 1981;
4 Raleigh, NC. New York, NY: Elsevier Biomedical; pp. 295-320. (Developments in toxicology and
5 environmental science: v. 10).
6
7 Nikula, K. J.; Snipes, M. B.; Barr, E. B.; Mauderly, J. L. (1991) Histopathology and lung tumor responses in
8 rats exposed to diesel exhaust or carbon black. In: Annual report of the Inhalation Toxicology Research
9 Institute operated for the United States Department of Energy by the Lovelace Biomedical and
10 Environmental Research Institute: October 1, 1990 though September 30, 1991. Albuquerque, NM:
11 Inhalation Toxicology Research Institute, Lovelace biomedical and Environmental Research Institute;
12 pp. 87-129; report no. LMF-134.
13
14 Nikula, K. J.; Snipes, M. B.; Barr, E. B.; Griffith, W. C.; Henderson, R. F.; Mauderly, J. L. (1994)
15 Influence of particle-associated organic compounds on the carcinogenicity of diesel exhaust.
16 In: Mohr, U.; Dungworth, D. L.; Mauderly, J. L.; Oberdorster, G., eds. Toxic and carcinogenic
17 effects of solid particles in the respiratory tract: [proceedings of the 4th international inhalation
18 symposium]; March 1993; Hannover, Germany. Washington, DC: International Life Sciences
19 Institute Press; pp. 565-568.
20
21 Orthoefer, J. G.; Moore, W.; Kraemer, D. (1981) Carcinogenicity of diesel exhaust as tested in strain A mice.
22 Environ. Int. 5: 461-471.
23
24 Pepelko, W. E.; Peirano, W. B. (1983) Health effects of exposure to diesel engine emissions: a summary of
25 animal studies conducted by the U.S. Environmental Protection Agency's Health Effects Research
26 Laboratories at Cincinnati, Ohio. J. Am. Coll. Toxicol. 2: 253-306.
27
28 Shefner, A. M.; Collins, B. R.; Dooley, L.; Fiks, A.; Graf, J. L.; Preache, M. M. (1982) Respiratory
29 carcinogenicity of diesel fuel emissions interim results. In: Lewtas, J., ed. Toxicological effects of
30 emissions from diesel engines: proceedings of the Environmental Protection Agency 1981 diesel
31 emissions symposium; October 1981; Raleigh, NC. New York, NY: Elsevier Biomedical; pp. 329-350.
32 (Developments in toxicology and environmental science: v. 10).
33
34 Stober, W. (1986) Experimental induction of tumors in hamsters, mice and rats after long-term inhalation of
35 filtered and unfiltered diesel engine exhaust. In: Ishinishi, N.; Koizumi, A.; McClellan, R. O.;
36 Stober, W., eds. Carcinogenic and mutagenic effects of diesel engine exhaust: proceedings of the
37 international satellite symposium on lexicological effects of emissions from diesel engines; July; Tsukuba
38 Science City, Japan. Amsterdam, The Netherlands: Elsevier Science Publishers B. V.; pp. 421-439.
39 (Developments in toxicology and environmental science: v. 13).
40
41 Takaki, Y.; Kitamura, S.; Kuwabara, N.; Fukuda, Y. (1989) Long-term inhalation studies of exhaust from diesel
42 engine in F-344 rats: the quantitative relationship between pulmonary hyperplasia and anthracosis. Exp.
43 Pathol. 37: 56-61.
44
45 Takemoto, K.; Yoshimura, H.; Katayama, H. (1986) Effects of chronic inhalation exposure to diesel exhaust on
46 the development of lung tumors in di-isopropanol-nitrosamine-treated F344 rats and newborn C57BL and
47 ICR mice. In: Ishinishi, N.; Koizumi, A.; McClellan, R. O.; Stober, W., eds. Carcinogenic and
48 mutagenic effects of diesel engine exhaust: proceedings of the international satellite symposium on
49 lexicological effects of emissions from diesel engines; July; Tsukuba Science City, Japan. Amsterdam,
50 Holland: Elsevier Science Publishers B. V.; pp. 311-327. (Development in toxicology and environmenlal
51 science: v. 13).
52
53
December 1994 7-48 DRAFT-DO NOT QUOTE OR CITE
-------
1 Takemoto, K.; Katayama, H.; Kuwabara, T.; Hasumi, M. (1988) Carcinogenicity by subcutaneous administration
2 of diesel paniculate extracts in mice. In: Diesel exhaust and health risks: results of the HERP studies.
3 Tsukuba, Ibaraki, Japan: Japan Automobile Research Institute, Inc., Research Committee for HERP
4 Studies; pp. 227-234.
5
6 Vuk, C. T.; Jones, M. A.; Johnson, J. H. (1976) The measurement and analysis of the physical character of
7 diesel paniculate emissions. SAE Trans. 85: 556.
8
9 White, H.; Vostal, J. J.; Kaplan, H. L.; MacKenzie, W. F. (1983) A long-term inhalation study evaluates the
10 pulmonary effects of diesel emissions [letter]. J. Appl. Toxicol. 3: 332.
11
12 Wong, D.; Mitchell, C. E.; Wolff, R. K.; Mauderly, J. L.; Jeffrey, A. M. (1986) Identification of DNA damage
13 as a result of exposure of rats to diesel engine exhaust. Carcinogenesis (London) 7: 1595-1597.
14
15 Zamora, P. 0.; Gregory, R. E.; Brooks, A. L. (1983) In vitro evaluation of the tumor-promoting potential of
16 diesel-exhaust-particle extracts. J. Toxicol. Environ. Health 11: 187-197.
December 1994 7.49 DRAFT-DO NOT QUOTE OR CITE
-------
i 8. EPIDEMIOLOGIC STUDIES OF THE
2 CARCINOGENICITY OF EXPOSURE TO
3 DIESEL EMISSIONS
4
5
6 8.1 INTRODUCTION
7 Emissions from diesel engine exhaust are made up of toxicants that include oxides of
8 nitrogen and sulfur, carbon monoxide, and particulate matter consisting of a carbon core with
9 many organic compounds, especially the polycyclic aromatic hydrocarbons adsorbed on the
10 surface. Diesel engine exhaust contains about 100 times more particulate matter than
11 gasoline engine exhaust.
12 In this chapter, various mortality and morbidity studies of the health effects of exposure
13 to diesel engine emissions are reviewed. Although an attempt was made to cover all the
14 relevant studies, a number of studies are not included for several reasons. First, the change
15 from steam to diesel engines in locomotives began in 1935 and was about 95% complete by
16 1959 (Garshick et al., 1988). Diesel buses also were introduced about the same time.
17 Therefore, exposure to diesel exhaust was less common, and the follow-up period for studies
18 conducted prior to 1959 (Raffle, 1957; Kaplan, 1959) was not long enough to cover the long
19 latency period of lung cancer. The usefulness of the studies in evaluating the carcinogenicity
20 of diesel exhaust is greatly reduced; thus, these studies are not considered here.
21 Second, hypothesis-generating studies were excluded from this review because the
22 findings of such studies need subsequent confirmation by definitive studies (Silverman et al.,
23 1983; Schenker et al., 1984; Buiatti et al., 1985; Flodin et al., 1987; Siemiatycki et al.,
24 1988).
25 Third, studies in which exposure to diesel exhaust was uncertain or was defined as
26 motor exhaust (which includes both gasoline and diesel exhaust) were excluded from the
27 review as they would have contributed little to the evaluation of the carcinogenicity of diesel
28 exhaust (Waxweiler et al., 1973; Ahlberg et al., 1981; Stern et al., 1981; Vineis and
29 Magnani, 1985; Gustafsson et al., 1986; Silverman et al., 1986; Jensen et al., 1987; Garland
30 et al., 1988; Risch et al., 1988).
December 1994 8-1 DRAFT-DO NOT QUOTE OR CITE
-------
1 Fourth, the study by Coggon et al. (1984) was not included because the occupational
2 information abstracted from death certificates had not been validated; this would have
3 resulted in limited information.
4 Three types of studies of the health effects of exposure to diesel engine emissions are
5 reviewed in this chapter: (1) cohort studies, (2) case-control studies of lung cancer, and
6 (3) case-control studies of bladder cancer. In the cohort studies, the cohorts of heavy
7 construction equipment operators, railroad and locomotive workers, and bus garage
8 employees were studied retrospectively to determine increased mortality and morbidity
9 resulting from exposures to varying levels of diesel emissions in the workplace. A total of
10 seven cohort mortality, eight lung cancer case-control, and seven bladder cancer case-control
11 studies are considered in this section.
12
13
14 8.2 COHORT STUDIES
15 8.2.1 Waller (1981): Trends in Lung Cancer in London in Relation to
16 Exposure to Diesel Fumes
17 A retrospective mortality study of a cohort of London transport workers was conducted
18 to determine if there was an excess of deaths from lung cancer that could be attributed to
19 diesel exhaust exposure. Nearly 20,000 male employees aged 45 to 64 were followed for the
20 25-year period between 1950 and 1974, constituting a total of 420,700 man-years at risk.
21 These were distributed among five job categories: drivers, garage engineers, conductors,
22 motormen or guards, and engineers (works). Most employees lived in the greater London
23 area. Lung cancer cases occurring in this cohort were ascertained only from death
24 certificates of individuals who died while still employed, or if retired, following diagnosis.
25 Expected death rates were calculated by applying greater London death rates to the
26 population at risk within each job category. Data were calculated in 5-year periods and
27 5-year age ranges, finally combining the results to obtain the total expected deaths in the
28 required age range of 45 to 64 for the calendar period from 1950 to 1974. A total of
29 667 cases of lung cancer was reported, compared with 849 expected, to give a mortality ratio
30 of 79%. In each of the five job categories, the observed numbers were below those
31 expected. Engineers in garages had the highest mortality ratio (90%) but this did not differ
December 1994 8-2 DRAFT-DO NOT QUOTE OR CITE
-------
1 significantly from the other job categories. Environmental sampling was done at one garage,
2 on 1 day in 1979, for benzo[a]pyrene concentrations and was compared with corresponding
3 values recorded in 1957. Concentrations of benzo[a]pyrene recorded in 1957 were at least
4 10 times greater than those measured in 1979.
5 This study has several methodologic limitations. The lung cancer deaths ascertained for
6 the study were those that occurred while the worker was employed (the worker either died of
7 lung cancer or retired after lung cancer was diagnosed). Although man-years at risk were
8 based on the entire cohort, no attempt was made to trace or evaluate the individuals who had
9 resigned from the London transport company for any other reason. Hence, the information
10 on resignees who may have had significant exposure to diesel exhaust, and lung cancer
11 deaths among them, was not available for analysis. This fact may have lead to a dilution
12 effect, resulting in underascertainment of observed lung cancer deaths and underestimation of
13 mortality ratios. Eligibility criteria for inclusion in the cohort, such as starting date and
14 length of service with the company, were not specified. Because an external comparison
15 group was used to obtain expected number of deaths, the resulting mortality ratios were less
16 than one; this may be a reflection of the "healthy worker effect". Investigators also did not
17 categorize the five job categories by levels of diesel exhaust exposure nor did they use an
18 internal comparison group to derive risk estimates.
19 The age range considered for this study was limited (45 to 64 years of age) for the
20 period between 1950 and 1964. It is not clear whether this age range was applied to
21 calendar year 1950 or 1964 or at the mid-point of this 25-year follow-up period.
22 No analyses were presented either by latency or by duration of employment (surrogate for
23 exposure). The environmental survey based on benzo[a]pyrene concentrations suggests that
24 the cohort in their earlier years was exposed to much higher concentrations of environmental
25 contaminants than current concentrations. It is not clear when the reduction in
26 benzo[a]pyrene concentration occurred since there are no environmental readings available
27 between 1957 and 1979. It is also important to note that the concentrations of
28 benzo[fl]pyrene inside the garage were not very different from those outside the garage in
29 1957, thus indicating that the exposure for garage workers was not much different than for
30 the general population. Lastly, no data were collected on smoking habits.
31
December 1994 8-3 DRAFT-DO NOT QUOTE OR CITE
-------
1 8.2.2 Howe et al. (1983): Cancer Mortality (1965 to 1977) in Relation to
2 Diesel Fume and Coal Exposure in a Cohort of Retired Railroad
3 Workers
4 This is a retrospective cohort study of the mortality experience of 43,826 male
5 pensioners of the Canadian National Railroad (CNR) between 1965 and 1977. Members of
6 this cohort consisted of male CNR pensioners who had retired before 1965 and who were
7 known to be alive at the start of that year, as well as those who retired between 1965 and
8 1977. The records were obtained from a computer file that is regularly updated and used by
9 the company for payment of pensions. To receive a pension, each pensioner must provide,
10 on a yearly basis, evidence to the effect that he is alive. Specific cause of death among
11 members of this cohort was ascertained by linking these records to the Canadian Mortality
12 Data Base, which contains records of all deaths registered in Canada since 1950. Of the
13 17,838 deaths among members of the cohort between 1965 and 1977, 16,812 (94.4%) were
14 successfully linked to a record on the mortality file. A random sample manual check on
15 unlinked data revealed that failure to link was mainly due to some missing information on the
16 death records.
17 Occupation at time of retirement was used by the Department of Industrial Relations to
18 classify workers into three diesel fume and coal dust exposure categories: (1) nonexposed,
19 (2) possibly exposed, and (3) probably exposed. Person-years of observation were calculated
20 and classified by age at observation in 5-year age groups (35 to 39, 40 to 44, . . ., 80 to 84,
21 and >85 years). The observed deaths were classified by age at death for different cancers,
22 for all cancers combined, and for all causes of death combined. Standard mortality ratios
23 (SMRs) were then calculated using rates of the Canadian population for the period between
24 1965 and 1977.
25 Both total mortality (SMR = 95, p < 0.001) and all cancer deaths (SMR = 99,
26 p > 0.05) were close to that expected for the entire cohort. Analysis by exposure to diesel
27 fume levels in the three categories (nonexposed, possibly exposed, and probably exposed)
28 revealed an increased relative risk for lung cancer among workers with increasing exposure
29 to diesel fumes. The relative risk for nonexposed workers was presumed to be 1.0; for those
30 possibly exposed, the relative risk was elevated to 1.2, which was statistically significant
31 (p = 0.013); and, for those probably exposed, it was elevated to 1.35, which was
32 statistically highly significant (p = 0.001). The corresponding rates for exposure to varying
December 1994 g_4 DRAFT-DO NOT QUOTE OR CITE
-------
1 levels of coal dust were very similar at 1.00, 1.21 (p = 0.012), and 1.35 (p = 0.001),
2 respectively. The trend tests were highly significant for both exposures (p < 0.001).
3 Analysis performed after the exclusion of individuals who worked in the maintenance of
4 steam engines, hence exposed to high levels of asbestos, yielded the risk of lung cancer to be
5 1.00, 1.21, and 1.33 for the nonexposed, possibly exposed, and probably exposed to diesel
6 exhaust, respectively, with a highly significant trend (p < 0.001).
7 An analysis done on individuals who retired prior to 1950 showed the relative risk of
8 lung cancer among nonexposed, possibly exposed, and probably exposed to be 1.00, 0.70,
9 and 0.44, respectively, based on less than 15 deaths in each category. A similar analysis of
10 individuals who retired after 1950 found the results in the same categories to be 1.00, 1.23,
11 and 1.40, respectively. Although retirement prior to 1950 indicated exposure to coal dust
12 alone, retirement after 1950 shows the results of mixed exposure to coal dust and diesel
13 fumes. As there was considerable overlap between occupations involving probable exposure
14 to diesel fumes and probable exposure to coal dust and as most members of the cohort were
15 employed during the years in which the transition from coal to diesel occurred, it was
16 difficult to distinguish whether lung cancer was associated with exposure to coal dust or
17 diesel fumes or a mixture of both.
18 Although this study showed a highly significant dose-response relationship between
19 diesel fumes and lung cancer, it has some methodological limitations. There were concurrent
20 exposures to both diesel fumes and coal dust during the transition period; therefore,
21 misclassification of exposure may have occurred, because only occupation at retirement was
22 available for analysis. It is possible that the elevated response observed for lung cancer was
23 due to the combined effects of exposure to both coal dust and diesel fumes and not just one
24 or the other. However, it should be noted that so far coal dust has not been demonstrated to
25 be a pulmonary carcinogen in studies of coal miners. No information was provided on
26 duration of employment in either diesel work or the coal dust-related jobs for other than
27 those jobs held at retirement. Therefore, it was not possible to evaluate whether this
28 omission would have led to an under- or overestimate of the true relative risk. Furthermore,
29 a lack of information on potential confounders such as smoking makes the interpretation of
30 the excess risk of lung cancer even more difficult. Information on cause of death was
December 1994 8-5 DRAFT-DO NOT QUOTE OR CITE
-------
1 acquired from the mortality data linkage. There is a possibility that the cause of death may
2 have been misclassified because of miscoding of the underlying cause of death.
3
4 8.2.3 Rushton et al. (1983): Epidemiological Survey of Maintenance
5 Workers in the London Transport Executive Bus Garages and
6 Chiswick Works
7 This is a retrospective mortality cohort study of male maintenance workers employed
8 for at least 1 continuous year between January 1, 1967, and December 31, 1975, at
9 71 London transport bus garages (also known as rolling stock) and at Chiswick Works. For
10 all men, the following information was obtained from computer listings: surname with
11 initials, date of birth, date of joining company, last or present jobs, and location of work.
12 For those individuals who left their job, date of and reason for leaving were also obtained.
13 For those who died in service or after retirement and for men who had resigned, full name
14 and last known address were obtained from an alphabetical card index in the personnel
15 department. Additional tracing of individuals who had left was carried out through social
16 security records. The area of their residence was assumed to be close to their work;
17 therefore their place of work was coded as their residence. There were 100 different job
18 titles that were coded into 20 broader groups. These 20 groups were not ranked for diesel
19 exhaust exposure though. The reason for leaving was coded as died in service, retired, or
20 other. The underlying cause of death was coded using the eighth revision of the International
21 Classification of Diseases (ICD). Person-years were calculated from date of birth and dates
22 of entry to and exit from the study using the man-years computer language program. These
23 were then subdivided into 5-year age and calendar period groups. The expected number of
24 deaths was calculated by applying the 5-year age and calendar period death rates of the
25 comparison population to the person-years of corresponding groups. The mortality
26 experience of the male population in England and Wales was used as the comparison
27 population. Significance values were calculated for the difference between the observed and
28 expected deaths, assuming a Poisson distribution.
29 The number of person-years of observation totaled 50,008 and was contributed by
30 8,490 individuals in the study with a mean follow-up of 5.9 years. Only 2.2% (194) of the
31 men were not traced. Observed deaths from all causes were significantly lower than
32 expected (observed = 495, p < 0.001). The observed deaths from all neoplasms and cancer
December 1994 8-6 DRAFT-DO NOT QUOTE OR CITE
-------
1 of the lung were approximately the same as those expected. The only significant excess
2 observed for cancer of the liver and gall bladder at Chiswick Works was based on four
3 deaths (p < 0.05). A few job groups showed a significant excess of risks for various
4 cancers. Cancer of the lung was elevated in the general hand category (SMR =133,
5 observed = 48, p < 0.03). Cancer of the liver and gall bladder (observed = 2, p < 0.01)
6 and cancer of the bladder (observed = 3, p < 0.01) were significant in the job categories of
7 inspector and progress hand, respectively. Welders showed an excess of lung cancer
8 (observed = 3, p < 0.05), bus mechanics had an excess of cancer of the brain
9 (observed = 4, p < 0.04), and painters had an excess of bladder cancer (observed = 2,
10 p < 0.02). All the excess deaths observed for the various job groups, except for the general
11 hand category, were based on very small numbers (usually smaller than five) and merited
12 cautious interpretation. Although the lung cancer excess in the general hand category was
13 based on 48 cases (SMR = 133), given the fact that there was no adjustment for confounding
14 variables such as smoking, the results should be interpreted cautiously.
15 This mortality study of London transport maintenance workers did not demonstrate any
16 cancer excesses based on a large number of cases; this needs further exploration. Its
17 limitations include the small sample size, short duration of follow-up (average of only
18 6 years), and lack of sufficient latency period, make this study inadequate to draw any
19 conclusions. The number of deaths by different causes and among the various job groups
20 was too small to allow any meaningful conclusions. Details of work history were not
21 obtained to permit any analysis by diesel exhaust exposure. Death information was
22 ascertained from death certificates with inherent problems of inaccuracy, misdiagnosis, and
23 errors in coding, and it was not known whether a trained nosologist coded the death
24 certificates. No adjustments were made for the confounding effects of smoking and
25 socioeconomic factors.
26
27 8.2.4 Wong et al. (1985): Mortality Among Members of a Heavy
28 Construction Operators Union with Potential Exposure to Diesel
29 Exhaust Emissions
30 This is a retrospective mortality study conducted on a cohort of 34,156 male members
31 of a heavy construction equipment operators union with potential exposure to diesel exhaust
32 emissions. Study cohort members were identified from records maintained at Operating
December 1994 8-7 DRAFT-DO NOT QUOTE OR CITE
-------
1 Engineers' Local Union No. 3-3A in San Francisco, CA. This union has maintained both
2 work and death records on all its members since 1964. Individuals with at least 1 year of
3 membership in this union between January 1, 1964, and December 31, 1978, were included
4 in the study. Work histories of the cohort were obtained from job dispatch computer tapes.
5 The study follow-up period was from January 1964 to December 1978. Death information
6 was obtained from a trust fund, which provided information on retirement dates, vital status,
7 and date of death for those who were entitled to retirement and death benefits.
8 Approximately 50% of the cohort had been union members for less than 15 years, whereas
9 the other 50% had been union members for 15 years or more. The average duration of
10 membership was 15 years. As of December 31, 1978, 29,046 (85%) cohort members were
11 alive, 3,345 (9.8%) were dead, and 1,765 (5.2%) remained untraced. Vital status of
12 10,505 members who had left the union as of December 31, 1978, were ascertained from the
13 Social Security Administration. Death certificates were obtained from appropriate state
14 health departments. Altogether, 3,243 deaths (for whom death certificates were available) in
15 the cohort were coded using the seventh revision of the ICD. Death certificates could not be
16 obtained for 102 individuals, only the date of death was available; these individuals were
17 included in the calculation of the SMR for all causes of death but were deleted from the
18 cause-specific SMR analyses. Expected deaths and SMRs were calculated using the U.S.
19 national age-sex-race cause-specific mortality rates for 5-year time periods between 1964 and
20 1978. The entire cohort population contributed to 372,525.6 person-years in this 5-year
21 study period.
22 A total of 3,345 deaths was observed, compared with 4,109 expected. The
23 corresponding SMR for all causes was 81.4 (p = 0.01), which confirmed the "healthy
24 worker effect". A total of 817 deaths was attributed to malignant neoplasms, slightly fewer
25 than the 878.34 expected based on U.S. white male cancer mortality rates (SMR = 93.0,
26 p = 0.05). Mostly there were SMR deficits for cause-specific cancers, including lung cancer
27 for the entire cohort (SMR = 98.6, observed = 309). The only significant excess SMR was
28 observed for cancer of the liver (SMR = 166.7, observed = 23, p < 0.05).
29 Analysis by length of union membership as a surrogate of duration for potential
30 exposure showed statistically significant increases in SMRs of cancer of the liver
31 (SMR = 424, p < 0.01) in the 10- to 14-year membership group and of the stomach
December 1994 8-8 DRAFT-DO NOT QUOTE OR CITE
-------
1 (SMR = 248, p < 0.05) in the 5- to 9-year membership group. No cancer excesses were
2 observed in the 15- to 19-year and 20+-year membership groups. Although the SMR for
3 cancer of the lung had a statistically significant deficit in the less than 5-year duration group,
4 it showed a positive trend with increasing length of membership, which leveled off after
5 10 to 14 years.
6 Cause-specific mortality analysis by latency period showed a positive trend for SMRs of
7 all causes of death, although all of them were statistically significant deficits, reflecting the
8 diminishing "healthy worker effect". This analysis also demonstrated a statistically
9 significant SMR excess for cancer of the liver (10- to 19-year group, SMR = 257.9). The
10 SMR for cancer of the lung showed a statistically significant deficit for a < 10-year latency,
11 but showed a definite positive trend with increasing latency.
12 In addition to these analyses of the entire cohort, similar analyses were carried out in
13 various subcohorts. Analyses of retirees, 6,678 individuals contributing to 32,670.1 person-
14 years, showed statistically significant increases (p < 0.01) in SMRs for all cancers
15 (SMR = 145.3, observed = 389), all causes of death (SMR = 114.5, observed = 1,345),
16 cancer of the digestive system (SMR = 142.4, observed = 103), cancer of the large intestine
17 (SMR = 181.8, observed = 46), cancer of the respiratory system (162.4, observed = 161),
18 cancer of the lung (SMR = 164.1, observed = 155), emphysema (SMR = 277.3,
19 observed = 75), and cirrhosis of the liver (SMR = 173.5, observed = 38). The other two
20 significant excesses (p < 0.01) were for lymphosarcoma and reticulosarcoma
21 (SMR = 231.2, observed = 10) and nonmalignant respiratory diseases (SMR = 129.0,
22 observed = 112). Further analysis of the 4,075 retirees (18,677.8 person-years) who retired
23 at age 65 or who retired earlier but had reached the age of 65, revealed statistically
24 significant SMR increases for all cancers (SMR = 114.7, observed = 224, p = 0.05),
25 cancer of the lung (SMR = 130, observed = 86, p < 0.05), and lymphosarcoma and
26 reticulosarcoma (SMR = 266.5, observed = 8, p < 0.05).
27 To analyze cause-specific mortality by job held (potential exposure to diesel exhaust
28 emissions), 20 functional job titles were used, which were further grouped into three
29 potential categories: (1) high exposure, (2) low exposure, and (3) unknown exposure.
30 A person was classified in a job title if he ever worked on that job. Based on this
31 classification system, if a person had ever worked in a high-exposure job title he was
December 1994 8-9 DRAFT-DO NOT QUOTE OR CITE
-------
1 included in that group, even though he may have worked for a longer time in a low-exposure
2 group or in an unknown exposure group. Information on length of work in any particular
3 job, hence indirect information on potential length of exposure, was not available either.
4 For the high-exposure group a statistically significant excess was observed for bulldozer
5 operators who had 15 to 19 years of membership and 20+ years of follow-up for cancer of
6 the lung (SMR = 343.4, p < 0.05). This excess was based on 5 out of 495 deaths observed
7 in this group of 6,712 individuals who contributed 80,327.6 person-years of observation.
8 The cause-specific mortality analysis in the low-exposure group revealed statistically
9 significant SMR excesses in individuals who had ever worked as engineers. These excesses
10 were for cancer of the large intestine (SMR = 807.2, observed = 3, p < 0.05) among those
11 with 15 to 19 years of membership and length of follow-up of at least 20 years, and cancer
12 of the liver (SMR = 871.9, observed = 3, p < 0.05) among those with 10 to 14 years of
13 membership and length of follow-up of 10 to 19 years. There were 7,032 individuals who
14 contributed to 78,402.9 person-years of observation in the low-exposure group.
15 For the unknown exposure group, a statistically significant SMR was observed for
16 motor vehicle accidents only (SMR = 173.3, observed = 21, p < 0.05). There were
17 3,656 individuals who contributed to 33,388.1 person-years of observation in this category.
18 No work histories were available for those who started their jobs before 1967 and for
19 those who held the same job prior to and after 1967. This constituted 9,707 individuals
20 (28% of the cohort) contributing to 104,447.5 person-years. Statistically significant SMR
21 excesses were observed for all cancers (SMR = 112, observed = 339, p < 0.05) and cancer
22 of the lung (SMR = 119.3, observed = 141, p < 0.01). A significant SMR elevation was
23 also observed for cancer of the stomach (SMR = 199.1, observed = 30, p < 0.01).
24 This study demonstrates a statistically significant excess for cancer of the liver, but also
25 shows statistically significant deficits in cancers of the large intestine and rectum. It may be,
26 as the authors suggested, that the liver cancer cases were actually cases resulting from
27 metastases from the large intestine and/or rectum, since tumors of these sites will frequently
28 metastasize to the liver. The excess in liver cancer mortality and the deficits in mortality
29 that are due to cancer of the large intestine and rectum could also, as the authors indicate, be
30 due to misclassification. Both possibilities have been considered by the investigators in their
31 discussion.
December 1994 8-10 DRAFT-DO NOT QUOTE OR CITE
-------
1 Cancer of the lung showed a positive trend with length of membership as well as with
2 latency, although none of the SMRs were statistically significant except for the workers
3 without any work histories. The individuals without any work histories may have been the
4 ones who were in their jobs for the longest period of time, because workers without job
5 histories included those who had the same job before and after 1967 and thus may have
6 worked 12 to 14 years or longer. If they had belonged to the category in which heavy
7 exposure to diesel exhaust emissions was very common for this prolonged time, then the
8 increase in lung cancer, as well as stomach cancer, might be linked to diesel exhaust.
9 Further information on those without work histories should be obtained if possible since such
10 information may be quite informative with regard to the evaluation of the carcinogenicity of
11 diesel exhaust.
12 The study design is adequate, covers about a 15-year observation period, has a large
13 enough population, and is appropriately analyzed; however, it has too many limitations to
14 permit any conclusions. First, no exposure histories are available. One has to make do with
15 job histories which provide limited information on exposure level. Any person who ever
16 worked at the job or any person working at the same job over any period of time are
17 included in the same category; this would have a dilution effect, since extremely variable
18 exposures were considered in the study. Second, the length of time worked in any particular
19 job is not available. Third, work histories were not available for 9,707 individuals who
20 contributed 104,447.5 person-years, a large proportion of the study cohort (28%). These
21 individuals happen to show the most evidence of a carcinogenic effect. Confounding by
22 alcohol consumption for cancer of the liver and smoking for emphysema and cancer of the
23 lung were not ruled out. Lastly, though 34,156 members were eligible for the study, the
24 vital status of 1,765 individuals was unknown. Nevertheless, they were still considered in
25 the denominator of all the analyses. The investigators fail to mention how the person-year
26 calculation for these individuals was handled. Also, some of the person-years might have
27 been overestimated, as people may have paid the dues for a particular year and then left
28 work. These two causes of overestimation of the denominator may have resulted in some or
29 all the SMRs to be underestimated.
30 As for the smoking survey, the investigators took a very small sample (133 out of
31 34,156, which was not even 1%). Of 133, only 107 (80%) participated. It was a systematic
December 1994 8-11 DRAFT-DO NOT QUOTE OR CITE
-------
1 sample, but the authors neglected to mention how the list was prepared for this systematic
2 sample. Hence, the sample may not be representative of the study population, and, with a
3 small sample size, the results are not generalizable. The questionnaire asked only for current
4 smoking history. No detailed history was obtained for the amount smoked or length of
5 smoking history, both of which have a bearing on emphysema as well as lung carcinoma.
6
7 8.2.5 Edling et al. (1987): Mortality Among Personnel Exposed to Diesel
8 Exhaust
9 This is a retrospective cohort mortality study of bus company employees, which
10 investigated a possible increased mortality in cardiovascular diseases and cancers from diesel
11 exhaust exposure. The cohort comprised all males employed at five different bus companies
12 in southeastern Sweden between 1950 and 1959. Using information from personnel
13 registers, individuals were classified into one or more categories and could have contributed
14 person-years at risk in more than one exposure category. The study period was from 1951 to
15 1983; information was collected from the National Death Registry, and copies of death
16 certificates were obtained from the National Bureau of Statistics. Workers who died after
17 age 79 were excluded from the study because diagnostic procedures were likely to be more
18 uncertain at higher ages (according to investigators). The cause-, sex-, and age-specific
19 national death rates in Sweden were applied to the 5-year age categories of person-years of
20 observation to determine expected deaths for all causes, malignant diseases, and
21 cardiovascular diseases. A Poisson distribution was used to calculate p-values and
22 confidence limits for the ratio of observed to expected deaths. The total cohort of 694 men
23 (after loss of 5 men to follow-up) was divided into three exposure categories: (1) clerks with
24 the lowest exposure, (2) bus drivers with moderate exposure, and (3) bus garage workers
25 with highest exposure.
26 The 694 men provided 20,304 person-years of observation with 195 deaths compared to
27 237 expected. A deficit in cancer deaths largely accounted for this lower than expected
28 mortality in the total cohort. Among subcohorts, no difference between observed and
29 expected deaths for total mortality, total cancers, or cardiovascular causes was observed for
30 clerks (lowest diesel exposure), bus drivers (moderate diesel exposure), and garage workers
December 1994 8-12 DRAFT-DO NOT QUOTE OR CITE
-------
1 (high diesel exposure). The risk ratios for all three categories were less than one except for
2 cardiovascular diseases among bus drivers, which was 1.1.
3 When the analysis was restricted to members who had at least a 10-year latency period
4 and either any exposure or an exposure exceeding 10 years, similar results were obtained
5 with fewer neoplasms than expected, whereas cardiovascular diseases showed risk around or
6 slightly above unity.
7 Five lung cancer deaths were observed among bus drivers who had moderate diesel
8 exhaust exposure while 7.2 were expected. The only other lung cancer death was observed
9 among bus garage workers who had the highest diesel exhaust exposure. The small size of
10 the cohort and poor data on diesel exhaust exposure are among the major limitations of this
11 study. Although lifetime occupational histories were available, no industrial hygiene data
12 were presented to validate the classification of workers into low, moderate, and high
13 exposure to diesel exhaust based on job title. The power of the present study was estimated
14 to be 80% to detect a relative risk of 1.2 for cardiovascular diseases and 1.4 for cancers, but
15 for specific cancer sites, the power was much lower than this. No information was available
16 on confounding effects of smoking and asbestos exposure at the work sites.
17
18 8.2.6 Boffetta and Stellman (1988): Diesel Exhaust Exposure and
19 Mortality Among Males in the American Cancer Society Prospective
20 Study
21 Boffetta and Stellman conducted a mortality analysis of 46,981 males whose vital status
22 was known at the end of the first 2 years of follow-up. The analysis was restricted to males
23 aged 40 to 79 years in 1982 who enrolled in the American Cancer Society's prospective
24 mortality study of cancer. Mortality was analyzed in relation to exposure to diesel exhaust
25 exposure and to employment in selected occupations related to diesel exhaust exposure.
26 In 1982, more than 77,000 American Cancer Society volunteers enrolled over 1.2 million
27 men and women from all 50 states, the District of Columbia, and Puerto Rico in a long-term
28 cohort study, the Cancer Prevention Study II (CPS-II). Enrollees were usually friends,
29 neighbors, or relatives of the volunteers; enrollment was by family groups with at least one
30 person hi the household 45 years of age or older. Subjects were asked to fill out a four-page
31 confidential questionnaire and return it in a sealed envelope. The questionnaire included
32 history of cancer and other diseases; use of medications and vitamins; menstrual and
December 1994 8-13 DRAFT-DO NOT QUOTE OR CITE
-------
1 reproductive history; occupational history; and information on diet, drinking, smoking, and
2 other habits. The questionnaire also included three questions on occupation: (1) current
3 occupation, (2) last occupation, if retired, and (3) job held for the longest period of time, if
4 different from the other two. Occupations were coded to an ad hoc two-digit classification in
5 70 categories. Exposures at work or in daily life to any of the 12 groups of substances were
6 also ascertained. These included diesel engine exhausts, asbestos, chemicals/acids/solvents,
7 dyes, formaldehyde, coal or stone dusts, and gasoline exhausts. Volunteers checked whether
8 their enrollees were alive or dead and recorded the date and place of all deaths every other
9 year during the study. Death certificates were then obtained from state health departments
10 and coded according to a system based on the ninth revision of the ICD by a trained
11 nosologist.
12 The data were analyzed to determine the mortality for all causes and lung cancer in
13 relation to diesel exhaust exposure, mortality for all causes and lung cancer and employment
14 in selected occupations with high diesel exhaust exposure, and mortality from other causes in
15 relation to diesel exhaust exposure. The incidence-density ratio was used as a measure of
16 association, and test-based confidence limits were calculated by the Miettinen method. For
17 stratified analysis, the Mantel-Haenszel method was used for testing linear trends. Data on
18 476,648 subjects comprising 939,817 person-years of risk were available for analysis. Three
19 percent of the subjects (14,667) had not given any smoking history, and 20% (98,026) of
20 them did not give information on diesel exhaust exposure and were therefore excluded from
21 the main diesel exhaust analysis. Among individuals who had provided diesel exhaust
22 exposure history, 62,800 were exposed and 307,143 were not exposed. Comparison of the
23 population with known information on deisel exhaust exposure with the excluded population
24 with no information on deisel exhaust exposure showed that the mean ages were 54.7 and
25 57.7 years, the nonsmokers were 72.4 and 73.2%, and the total mortality rates per 1,000 per
26 year were 23.0 and 28.8 %, respectively.
27 The all-cause mortality was elevated among railroad workers (relative risk [RR] =
28 1.43, 95% confidence interval [CI] = 1.2, 1.72), heavy equipment operators (RR =1.7,
29 95% CI = 1.19, 2.44), miners (RR = 1.34, 95% CI = 1.06, 1.68), and truck drivers
30 (RR = 1.19, 95% CI = 1.07, 1.31). For lung cancer mortality the risks were significantly
31 elevated for miners (RR = 2.67, 95% CI = 1.63, 4.37) and heavy equipment operators
December 1994 8-14 DRAFT-DO NOT QUOTE OR CITE
-------
1 (RR = 2.60, 95% CI = 1.12, 6.06). Risks were also elevated but not significantly for
2 railroad workers (RR = 1.59, 95% CI = 0.94, 2.69) and truck drivers (RR = 1.24,
3 95% CI = 0.93, 1.66). These risks were calculated according to the Mantel-Haenszel
4 method, controlling for age and smoking. Although the relative risk was nonsignificant for
5 truck drivers, a small dose-response effect was observed when duration of diesel exhaust
6 exposure for them was examined. For drivers who worked for 1 to 15 years, the relative
7 risk was 0.87, while for drivers who worked for more than 16 years the relative risk was
8 1.33 (95% CI = 0.64, 2.75). Relative risks for lung cancer were not presented for other
9 occupations. Mortality analysis for other causes and diesel exhaust exposure showed a
10 significant excess of deaths (p < 0.05) in the following categories: cerebrovascular disease,
11 arteriosclerosis, pneumonia, influenza, cirrhosis of the liver, and accidents.
12 The two main methodologic concerns in this study are the representativeness of the
13 study population and the quality of information on exposure. The sample, though very large,
14 was comprised of volunteers. Thus, the cohort was healthier and less frequently exposed to
15 important risk factors such as smoking and alcohol. Self-administered questionnaires were
16 used to obtain data on occupation and diesel exhaust exposure. None of this information was
17 validated. Nearly 20% of the individuals had an unknown exposure status to diesel exhaust,
18 and they experienced a higher mortality for all causes and lung cancer than both the diesel
19 exhaust exposed and unexposed groups. This could have introduced a substantial bias in the
20 estimate of the association. Although only 0.8% of the subjects were lost to follow-up, the
/
21 use of death certificates alone as a source of medical information poses problems in accuracy
22 and coding. But the authors report that cancer deaths are routinely checked by histological
23 confirmation from physicians or cancer registries. Given the fact that all diesel exhaust
24 exposure occupations, such as heavy equipment operators, truck drivers, and railroad
25 workers showed elevated lung cancer risk, this study is suggestive of a causal association
26 between the two.
27
28 8.2.7 Garshick et al. (1988): A Retrospective Cohort Study of Lung
29 Cancer and Diesel Exhaust Exposure in Railroad Workers
30 An earlier case-control study of lung cancer and diesel exhaust exposure in U.S.
31 railroad workers by these investigators had demonstrated a relative odds of
December 1994 g_15 DRAFT-DO NOT QUOTE OR CITE
-------
1 1.41 (95% CI = 1.06, 1.88) for lung cancer with 20 years of work in jobs with diesel
2 exhaust exposure. To confirm these results, a large retrospective cohort mortality study was
3 conducted by the same investigators. Data sources for the study were the work records of
4 the U.S. Railroad Retirement Board (RRB). The cohort was selected based on job titles in
5 1959, which was the year by which 95% of the locomotives in the United States were diesel
6 powered. Diesel exhaust exposure was considered to be a dichotomous variable depending
7 on yearly job codes between 1959 and death or retirement through 1980. Industrial hygiene
8 evaluations and descriptions of job activities were used to classify jobs as exposed or
9 unexposed to diesel emissions. A questionnaire survey of 534 workers at one of the
10 railroads where workers were asked to indicate the amount of time spent in railroad
11 locations, either near or away from sources of diesel exhaust, was used to validate this
12 classification. Workers selected for this survey were actively employed at the tune of the
13 survey, 40 to 64 years of age, who started work between 1939 and 1949, in the job codes
14 sampled in 1959, and were eligible for railroad benefits. To qualify for benefits, a worker
15 must have 10 years or more of service with the railroad and should not have worked for
16 more than 2 years in a nonrailroad job after leaving railroad work. Workers with recognized
17 asbestos exposure, such as repair of asbestos-insulated steam locomotive boilers, passenger
18 cars, and steam pipes, or railroad building construction and repairs were excluded from the
19 job categories selected for study. However, a few jobs with some potential for asbestos
20 exposure were included in the cohort, and the analysis was done both ways with and without
21 them.
22 The death certificates for all subjects identified in 1959 and reported by the RRB to
23 have died through 1980 were searched. Twenty-five percent of them were obtained from the
24 RRB and the remainder from the appropriate state departments of health. Coding of cause of
25 death was done without knowledge of exposure history, according to the eighth revision of
26 the ICD. If the underlying cause of death was not lung cancer, but was mentioned on the
27 death certificate, it was assigned as a secondary cause of death, so that the ascertainment of
28 all cases was complete. Workers not reported by the RRB to have died by December 31,
29 1980, were considered to be alive. Deceased workers for whom death certificates had not
30 been obtained, or if obtained did not indicate cause of death, were assumed to have died of
31 unknown causes.
December 1994 8-16 DRAFT-DO NOT QUOTE OR CITE
-------
1 Proportional hazard models were fitted that provided estimates of relative risk for death
2 caused by lung cancer using the partial likelihood method described by Cox, and 95%
3 confidence intervals were constructed using the asymptotic normality of the estimated
4 regression coefficients of the proportional hazards model. Exposure was analyzed by diesel
5 exhaust-exposed jobs in 1959 and by cumulative number of years of diesel exhaust exposure
6 through 1980. Directly standardized rate ratios for deaths from lung cancer were calculated
7 for diesel exhaust exposed compared with unexposed for each 5-year age group in 1959.
8 The standardized rates were based on the overall 5-year person-year time distribution of
9 individuals in each age group starting in 1959. The only exception to this was between 1979
10 and 1980, when a 2-year person-year distribution was used. The Mantel-Haenszel analogue
11 for person-year data was used to calculate 95% confidence intervals for the standardized rate
12 ratios.
13 The cohort consisted of 55,407 workers, of which 19,396 had died by the end of 1980.
14 Death certificates were not available for 11.7% of all deaths. Of the 17,120 deaths for
15 whom death certificates were obtained, 48.4% were attributable to diseases of the circulatory
16 system, whereas 21% were attributable to all neoplasms. Of all neoplasms, 8.7%
17 (1,694 deaths) were due to lung cancer. A higher proportion of workers in the younger age
18 groups, mainly brakemen and conductors, were exposed to diesel exhaust, while a higher age
19 of workers in the older age groups were potentially exposed to asbestos. In a proportional
20 hazards model, analyses by age in 1959 found a relative risk of 1.45 (95% CI = 1.11, 1.89)
21 among the age group 40 to 44 years and a relative risk of 1.33 (95% CI = 1.03, 1.73) for
22 the age group 45 to 49 years. Risk estimates in the older age groups 50 to 54, 55 to 59, and
23 60 to 64 years were 1.2, 1.18, and 0.99, respectively, and were not statistically significant.
24 The two youngest age groups in 1959 had workers with the highest prevalence and longest
25 duration of diesel exhaust exposure and lowest exposure to asbestos. When potential
26 asbestos exposure was considered as a confounding variable in a proportional hazards model,
27 the estimates of relative risk for asbestos exposure were all near null value and not
28 significant. Analysis of workers exposed to diesel exhaust in 1959 (n = 42,535), excluding
29 the workers with potential past exposure to asbestos, yielded relative risks of
30 1.57 (95% CI = 1.19, 2.06) and 1.34 (95% CI = 1.02, 1.76) in the 1959 age groups 40 to
31 44 years and 45 to 49 years. Directly standardized rate ratios were also calculated for each
December 1994 347 DRAFT-DO NOT QUOTE OR CITE
-------
1 1959 age group based on diesel exhaust exposure in 1959. The results obtained confirmed
2 those obtained by using the proportional hazards model.
3 Relative risk estimates were then obtained using duration of diesel exhaust exposure as
4 a surrogate for dose. In a model that used years of exposure up to and including exposure in
5 the year of death, no exposure duration-response relationship was obtained. When analysis
6 was done by disregarding exposure in the year of death and 4 years prior to death, the risk
7 of dying from lung cancer increased with the number of years worked in a diesel exhaust-
8 exposed job. In this analysis, exposure to diesel exhaust was analyzed by exposure duration
9 groups and in a model entering age in 1959 as a continuous variable. The workers with
10 greater than 15 years of exposure had a relative risk of lung cancer of 1.72 (95% CI = 1.27,
11 2.33). The risks for 1 to 4 years of cumulative exposure was 1.20 (95% CI = 1.01, 1.44),
12 for 5 to 9 years of cumulative exposure it was 1.24 (95% CI = 1.06, 1.44), and for 10 to
13 14 years of cumulative exposure it was 1.32 (95% CI = 1.13, 1.56). Directly standardized
14 rate ratios were also calculated for each 1959 age group based on diesel exposure in 1959.
15 The results obtained confirmed those obtained by using the proportional hazards model.
16 The results of this study, demonstrating a positive association between diesel exhaust
17 exposure and increased lung cancer, are consistent with the results of the case-control study
18 conducted by the same investigators in railroad workers dying of lung cancer from March
19 1981 through February 1982. This cohort study has addressed many of the weaknesses of
20 the other epidemiologic studies. The large sample size (60,000) allowed sufficient power to
21 detect small risks and also permitted the exclusion of workers with potential past exposure to
22 asbestos. The stability of job career paths in the cohort ensured that of the workers 40 to
23 44 years of age in 1959 classified as diesel exhaust- exposed, 94% of the cases were still in
24 diesel exhaust-exposed jobs 20 years later.
25 The main limitation of the study is the lack of quantitative data on exposure to diesel
26 exhaust. This is one of the few studies in which industrial hygiene measurements of diesel
27 exhaust were done. These measurements were correlated with job titles to divide the cohort
28 in dichotomous exposure groups of exposed and nonexposed. This may have lead to an
29 underestimation of the risk of lung cancer since exposed groups included individuals with low
30 to high exposure. The number of years exposed to diesel exhaust was used as a surrogate
31 for dose. The dose, based on duration of employment, may have been inaccurate because
December 1994 8-18 DRAFT-DO NOT QUOTE OR CITE
-------
1 individuals were working on steam or diesel locomotives during the transition period. If the
2 categories of exposure to diesel exhaust would have been set up as no, low, moderate, and
3 high exposure, the results would have been more meaningful and so would have been the
4 dose-response relationship. Another limitation of this study was the inability to examine the
5 effect of years of exposure and latency. No adjustment for smoking was done in this study.
6 However, an earlier case-control study done in the same cohort (Garshick et al., 1987)
7 showed no significant difference in the risk estimate after adjusting for smoking. Despite
8 these limitations, the results of this study demonstrate that occupational exposure to diesel
9 exhaust is associated with a modest risk (1.5) of lung cancer.
10 Table 8-1 summarizes the cohort studies.
11
12
13 8.3 CASE-CONTROL STUDIES OF LUNG CANCER
14 8.3.1 Williams et al. (1977): Associations of Cancer Site and Type with
15 Occupation and Industry from the Third National Cancer Survey
16 Interview
17 This paper reports findings of the analysis of the Third National Cancer Survey
18 (TNCS). The lifetime histories, occupations, and industries were studied for associations
19 with specific cancer sites and types after controlling for age, sex, race, education, use of
20 cigarettes or alcohol, and geographic location. Of 13,179 cancer patients, a 10% random
21 sample of all incident invasive cancers in eight areas, a total of 7,518 were successfully
22 interviewed in the 3 years surveyed by the TNCS. These comprised 57% of those eligible to
23 participate. The interview included items on use of tobacco and alcohol (by type, amount,
24 and duration), family income, patient education, and employment history. Actual
25 descriptions of the occupation and industry were recorded by interviewers and were coded
26 separately for main lifetime employment, recent employment, and other jobs held according
27 to the 1970 Census Coding Scheme. Occupations or industries were combined to form larger
28 groups. Coding of occupational and industrial labels in meaningful job categories was done
29 by one of the authors. Of the 3,539 interviewed males and 3,937 interviewed females,
30 95 and 84%, respectively, listed some main employment. The basic analysis consisted of an
31 intercancer comparison and involved comparing the proportions of specific main lifetime
December 1994 8-19 DRAFT-DO NOT QUOTE OR CITE
-------
D
o
I
i—'
VO
TABLE 8-1. EPIDEMIOLOGIC STUDIES OF THE HEALTH EFFECTS OF EXPOSURE TO DIESEL EXHAUST:
COHORT MORTALITY STUDIES
Authors
Population Studied
Diesel Exhaust Exposure
Assessment
Results
Limitations
Waller
(1981)
Approximately 20,000 male
London transportation workers
Aged 45 to 64 years
25 years follow-up (1950-1974)
Five job categories used
to define exposure
Environmental
benzo[a]pyrene
concentrations measured
in 1957 and 1979
SMR = 79 for lung
cancer for the total
cohort
SMRs for all five job
categories were less than
100 for lung cancer.
Exposure measurement of benzo[a]pyrene
showed very little difference between inside and
outside the garage.
Incomplete information on cohort members
No adjustment for confounding such as other
exposures, cigarette smoking, etc.
No latency analysis
oo
Howe et al. 43,826 Male pensioners of the
(1983) Canadian National Railway
Company
Mortality between 1965 and
1977 among these pensioners
was compared with mortality
of general Canadian
population.
Exposure groups
classified by a group
of experts based on
occupation at the time
of retirement
Three exposures groups:
Nonexposed
Possibly exposed
Probably exposed
RR = 1.2(p = 0.013)
RR = 1.3(p = 0.001)
for lung cancer for
possible and probable
exposure, respectively
A highly significant
dose-response relationship
demonstrated by trend
test (p < 0.001)
Incomplete exposure assessment due to lack of
lifetime occupational history
Mixed exposures to coal dust and diesel exhaust
No validation of method was used to categorize
exposure.
No data on smoking
No latency analysis
Rushton 8,490 Male London transport
et al. (1983) maintenance workers
Mortality of workers employed
for 1 continuous year between
January 1, 1967, and December
31, 1975, was compared with
mortality of general population
of England and Wales.
100 Different job titles
were grouped hi
20 broad categories.
The categories were not
ranked for diesel exhaust
exposure.
SMR = 133 (p < 0.03)
for lung cancer in the
general hand job group
Several other job
categories showed SS
increased SMRs for
several other sites based
on less than five cases.
Ill-defined diesel exhaust exposure without any
ranking
Average 6-year follow-up, (i.e. not enough time
for lung cancer latency)
No adjustment for confounders such as smoking
-------
TABLE 8-1 (cont'd). EPIDEMIOLOGIC STUDIES OF THE HEALTH EFFECTS OF EXPOSURE TO DIESEL
EXHAUST: COHORT MORTALITY STUDIES
Authors
Population Studied
Diesel Exhaust Exposure
Assessment
Results
Limitations
oo
Wong et al. 34,156 Male heavy construction
(1985) equipment operators
Members of the local union for
at least 1 year between
January 1, 1964, and
December 1, 1978
20 Functional job titles
grouped into three job
categories for potential
exposure
Exposure groups (high,
low, and unknown) based
on job description and
proximity to source of
diesel exhaust emissions
SMR = 166 (p < 0.05)
for liver cancer for total
cohort
SMR = 343 (observed =
5, p < 0.05) for lung
cancer for high exposure
bulldozer operators with
15-19 years of
membership, 20+ years
of follow-up
SMR = 119 (observed =
141, p < 0.01) for
workers with no work
histories
No validation of exposure categories, which
were based on surrogate information
Incomplete employment records
Employment history other than from the union
not available
No data on confounder such as other exposures,
smoking, etc.
Edling et al. 694 Male bus garage employees Three exposure groups
(1987)
Follow-up from 1951 through
1983
Mortality of these men was
compared with mortality of
general population of Sweden.
based on job titles:
High exposure, bus
garage workers
Intermediate exposure,
bus drivers
Low exposure, clerks
No SS differences were
observed between
observed and expected
for any cancers by
different exposure
groups.
Small sample size
No validation of exposure
No data on confounders such as other exposures,
smoking, etc.
-------
TABLE 8-1 (cont'd). EPIDEMIOLOGIC STUDIES OF THE HEALTH EFFECTS OF EXPOSURE TO DIESEL
EXHAUST: COHORT MORTALITY STUDIES
Authors
Population Studied
Diesel Exhaust Exposure
Assessment
Results
Limitations
Boffetta and 46,981 Male volunteers
Stellman enrolled in the American
(1988) Cancer Society's Prospective
Mortality Study of Cancer in
1982
Aged 40 to 79 years at
enrollment
First 2-year follow-up
Self-reported
occupations were coded
into 70 job categories.
Employment in high
diesel exhaust exposure
jobs were compared
with nonexposed jobs.
Total mortality (SS) elevated for railroad
workers, heavy equipment operators, miners,
and truck drivers
Lung cancer mortality (SS) elevated for miners
and heavy equipment operators
Lung cancer mortality (SNS) elevated among
railroad workers and truck drivers
Truck drivers also showed a small dose
response
Exposure information based on
self-reported occupation for which
no validation was done
Volunteer population, probably
healthy population
oo
tl>
to
Garshick 55,407 White male railroad
et al. (1988) workers
Aged 40 to 64 years in 1959
Started work 10-20 years
earlier than 1959
Industrial hygiene data RR = 1.45 (40-44 year age group)
correlated with job
titles to dichotomize
the jobs as "exposed"
or "not exposed"
RR = 1.33 (45-49 year age group)
BothSS
After exclusion of workers exposed to asbestos
RR = 1.57 (40-44 year age group)
RR = 1.34 (45-49 year age group)
BothSS
Dose response indicated by increasing lung
cancer risk with increasing cumulative
exposure
Years of exposure used as
surrogate for dose
Not possible to separate the effect
of time since first exposure and
duration of exposure
Abbreviations:
RR = Relative risk.
SMR = Standardized mortality ratio.
SNS = Statistically nonsignificant.
SS = Statistically significant.
-------
1 industries and occupations among patients with cancer at one site with those of patients
2 having cancer at other sites combined as a control group, and this was done using a series of
3 Mantel-Haenszel stratified contingency table analyses to yield odds ratios and chi-square
4 values. Odds ratios (ORs) were computed controlling for age, race, education, tobacco,
5 alcohol, and geographic location, and these were done separately for males and females.
6 A total of 432 and 128 lung cancers were present in males and females, respectively.
7 For males an excess risk of lung cancer was observed for the following main industrial
8 groups: mines (OR = 1.21), construction (OR = 1.24), transportation (OR = 1.17), utility
9 and sanitary services (OR = 2.79, p < 0.05), and professional (OR = 1.41). An excess of
10 bladder cancer was reported for the mining industry (OR = 1.61). For females, an excess
11 of lung cancer was detected for the transportation industry (OR = 1.96); finance and retail
12 industry (OR = 1.73); and the business, car repair, and miscellaneous service industry
13 (OR = 2.29). None of these excesses were statistically significant. All these odds ratios
14 were adjusted for age, race, education, tobacco, alcohol, and geographic location. The
15 transportation industry for males and females also showed a nonsignificant excess risk for
16 cancers of the liver and gall bladder ducts. When the analysis was done for specific lifetime
17 industries, the odds ratios for lung cancer in males was 1.40 for railroad workers and
18 1.34 for truck drivers. Both these excesses were statistically nonsignificant.
19 The strengths of the TNCS interview data set are its large size, histological
20 confirmation of nearly 95% of diagnoses, availability of information on occupation, and
21 details of confounding variables obtained by personal interview and ability to control for
22 them. Among its weaknesses are a 47% nonresponse rate and the fact that the population
23 surveyed came from predominantly urban areas and did not represent many industries. Also,
24 most of the associations observed did not achieve statistical significance because they were
25 based on small numbers of patients who had both specific cancers and specific types of
26 employment. The control group was the combined "other cancers" which may have diluted
27 the association since diesel exhaust is also suspected of being associated with bladder cancer,
28 and this category was included in the control group when the comparison was made with
29 lung cancer. The study presented several tables, but the total population in each table was
30 different and never added up to the initial number interviewed. The authors failed to explain
31 these omissions. Further, when multiple comparisons are made, some significant
December 1994 g_23 DRAFT-DO NOT QUOTE OR CITE
-------
1 associations arise by chance. This analysis does suggest an association with lung cancer for
2 three industries with potential diesel exhaust exposure. These were trucking, railroading,
3 and mining.
4
5 8.3.2 Hall and Wynder (1984): A Case-Control Study of Diesel Exhaust
6 Exposure and Lung Cancer
7 Hall and Wynder conducted a case-control study of 502 male lung cancer cases and
8 502 controls without tobacco-related diseases that examined an association between
9 occupational diesel exhaust exposure and lung cancer. Histologically confirmed primary lung
10 cancer patients who were 20 to 80 years old were ascertained from 18 participating hospitals
11 in six U.S. cities, 12 mo prior to the interview. Eligible controls, comprised patients at the
12 same hospitals without tobacco-related diseases, were matched to cases by age (±5 years),
13 race, hospital, and hospital room status. The number of male lung cancer cases interviewed
14 totaled 502, which was 64% of those who met the study criteria for eligibility. Of the
15 remaining 36%, 8% refused, 21% were too ill or had died, and 7% were unreliable.
16 Seventy-five percent of eligible controls completed interviews. Of these interviewed
17 controls, 49.9% were from the all cancers category, whereas 50.1% were from the all
18 noncancers category. All interviews were obtained in hospitals to gather detailed information
19 on smoking history, coffee consumption, artificial sweetener use, residential history, and
20 abbreviated medical history as well as standard demographic variables. Occupational
21 information was elicited by a question on the usual lifetime occupation and was coded by the
22 abbreviated list of the U.S. Bureau of Census Codes. The odds ratios were calculated to
23 evaluate the association between diesel exhaust exposure and risk of lung cancer incidence.
24 Summary odds ratios were computed by the Mantel-Haenszel method after adjusting for
25 potential confounding by age, smoking, and socioeconomic class. Two sided, 95%
26 confidence intervals were computed by Woolf's method. Occupational exposure to diesel
27 exhaust was defined by two criteria. First, occupational titles were coded "probably high
28 exposure" as defined by the industrial hygiene standards established for the various jobs.
29 The job titles included under this category were warehousemen, bus and truck drivers,
30 railroad workers, and heavy equipment operators and repairmen. The second method used
31 the National Institute for Occupational Safety and Health (NIOSH) criteria to analyze
December 1994 8-24 DRAFT-DO NOT QUOTE OR CITE
-------
1 occupations by diesel exposure. In this method, the estimated proportion of exposed workers
2 was computed for each occupational category by using the NIOSH estimates of the exposed
3 population as the numerator and the estimates of individuals employed in each occupational
4 category from the 1970 census as the denominator. Occupations estimated to have at least
5 20% of their employees exposed to diesel exhaust were defined as "high exposure," those
6 with 10 to 19% of their employees exposed were defined as "moderate exposure," and those
7 with less than 10% of their empolyees exposed were defined as "low exposure."
8 Cases and controls were compared with respect to exposure. The relative risk was
9 2.0 (95% CI = 1.2, 3.2) for those workers who were exposed to diesel exhaust versus those
10 who were not. The risk, however, decreased to a nonsignificant 1.4 when the data were
11 adjusted for smoking. Analysis by NIOSH criteria found a nonsignificant relative risk of
12 1.7 in the high exposure group. There were no significantly increased cancer risks by
13 occupation either by the first method or by the NIOSH method. In order to assess any
14 possible synergism between diesel exhaust exposure and smoking, the lung cancer risks were
15 calculated for different smoking categories. The relative risks were 1.46 among nonsmokers
16 and exsmokers, 0.82 among current smokers of <20 cigarettes/day, and 1.3 among current
17 smokers of 20+ cigarettes/day indicating a lack of synergistic effects.
18 The major strength of this study is the availability of a detailed smoking history for all
19 the study subjects. However, this is offset by the lack of diesel exhaust exposure
20 measurements, use of a poor surrogate for exposure, and lack of consideration of latency
21 period. Information was collected on only one major lifetime occupation, and it is likely that
22 those workers who had more than one major job may not have reported the occupation with
23 the heaviest diesel exhaust exposures. Further, occupational histories were obtained from
24 self reports and were not validated with work records. This could have resulted in recall
25 bias and misclassification of exposure status.
26
27 8.3.3 Damber and Larsson (1987): Occupation and Male Lung Cancer,
28 a Case-Control Study in Northern Sweden
29 A case-control study of lung cancer was conducted hi northern Sweden to determine the
30 occupational risk factors that could explain the large geographic variations of lung cancer
31 incidence in that country. The study region comprised the three northern-most counties of
December 1994 8-25 DRAFT-DO NOT QUOTE OR CITE
-------
1 Sweden with a total male population of about 390,000. The rural municipalities with 15 to
2 20% of the total population have forestry and agriculture as dominating industries and the
3 urban areas have a variety of industrial activities (mines, smelters, steel factories, paper
4 mills, and mechanical workshops). All male cases of lung cancer reported to the Swedish
5 Cancer Registry during the 6-year period between 1972 and 1977, who had died before the
6 start of the study, were selected. Of 604 eligible cases, 5 did not have microscopic
7 confirmation and in another 5 the diagnosis was doubtful, but these cases were included
8 nevertheless. Cases were classified as small carcinomas, squamous cell carcinomas,
9 adenocarcinomas, and other types. For each case a dead control was drawn from the
10 National Death Registry matched by sex, year of death, age, and municipality. Deaths in
11 controls classified as lung cancer and suicides were excluded. A living control matched to
12 the case by sex, year of birth, and municipality was also drawn from the National Population
13 Registry. Postal questionnaires were sent to close relatives of cases and dead controls, and
14 to living controls themselves to collect data on occupation, employment, and smoking habits.
15 Replies were received from 589 cases (98%) 582 surrogates of dead controls (96%), and
16 453 living controls (97%).
17 Occupational data were collected on occupations or employment held for at least 1 year
18 and included type of industry, company name, task, and duration of employment.
19 Supplementary telephone interviews were performed if occupational data were lacking for
20 any period between age 20 and time of diagnosis. Data analysis involved calculation of the
21 odds ratios by the exact method based on the hypergeometric distribution and the use of a
22 linear logistic regression model to adjust for the potential confounding effects of smoking.
23 Separate analyses were performed with dead and living controls, and on the whole there was
24 good agreement between the two control groups. A person who had been active for at least
25 1 year in a specific occupation was in the analysis assigned to that occupation.
26 Using dead controls, the odds ratios adjusted for smoking were 1.0 (95% CI = 0.7,
27 1.5) and 2.7 (95% CI = 1.0, 8.1) for professional drivers (>1 year of employment) and
28 underground miners (> 1 year of employment), respectively. For 20 or more years of
29 employment in those occupations, the odds ratios adjusted for smoking were
30 1.2 (95% CI = 0.6, 2.2) and 9.8 (95% CI = 1.5, 414). These were the only two
31 occupations listed with potential diesel exhaust exposure. An excess significant risk was
December 1994 8-26 DRAFT-DO NOT QUOTE OR CITE
-------
1 detected for copper smelter workers, plumbers, and electricians, as well as concrete and
2 asphalt workers. Occupational asbestos exposure was also associated with an elevated odds
3 ratio of 2.6 (95% CI = 1.6, 3.6) for >1 year of employment and 3.6 (95% CI = 1.9, 7.2)
4 for ^20 years of employment. All the odds ratios were calculated by adjusting for age,
5 smoking, and municipality. After comparison with the live controls, the odds ratios were
6 found to be lower than those observed with dead controls. None of the odds ratios were
7 statistically significant in this comparison.
8 This study did not detect any excess risk of lung cancer for professional drivers who,
9 among all the occupations listed, had the most potential for exposure to motor vehicle
10 exhaust. However, it is not known whether these drivers were exposed exclusively to
11 gasoline exhaust, diesel exhaust, or varying degrees of both. An excess risk was detected for
12 underground miners, but it is not known if this was due to diesel emissions from engines or
13 from radon daughters in poorly ventilated mines. Although a high response rate (98%) was
14 obtained by the postal questionnaires, the use of surrogate respondents is known to lead to
15 misclassification errors that can bias the odds ratio to one.
16
17 8.3.4 Lerchen et al. (1987): Lung Cancer and Occupation in New Mexico
18 This is a population-based case-control study conducted in New Mexico that examines
19 the association between occupation and occurrence of lung cancer in Hispanic and
20 non-Hispanic whites. Cases involved residents of New Mexico, 25 through 84 years of age
21 and diagnosed between January 1, 1980, and December 31, 1982, with primary lung cancer,
22 excluding bronchioalveolar carcinoma. Cases were ascertained through the New Mexico
23 Tumor Registry which is a member of the Surveillance Epidemiology and End Results
24 (SEER) Program of the National Cancer Institute. Controls were chosen by randomly
25 selecting residential telephone numbers, and for those over 65 years of age, from the Health
26 Care Financing Administration's roster of Medicare participants. They were frequency-
27 matched to cases for sex, ethnicity, and 10-year age category with a ratio of 1.5 controls per
28 case. The 506 cases (333 males and 173 females) and 771 controls (499 males and
29 272 females) were interviewed, with a nonresponse rate of 11% for cases. Next of kin
30 provided interviews for 50 and 43% of male and female cases, respectively. Among
31 controls, only 2% of the interviews were provided by next of kin for each sex. Data were
December 1994 8-27 DRAFT-DO NOT QUOTE OR CITE
-------
1 collected by personal interviews conducted by bilingual interviewers in the participants'
2 homes. A lifetime occupational history and a self-reported history of exposure to specific
3 agents were obtained for each job held for at least 6 mo since age 12. Questions were asked
4 about the title of the position, duties performed, location and nature of industry, and time at
5 each job title. A detailed smoking history was also obtained. The variables on occupational
6 exposures were coded according to the Standard Industrial Classification scheme by a single
7 person and reviewed by another. To test the hypothesis about the high-risk jobs for lung
8 cancer, an a priori listing of suspected occupations and industries was created by a two-step
9 process involving a literature review for implicated industries and occupations by the
10 principal investigator. The appropriate Standard Industrial Classification and Standard
11 Occupational Codes associated with job titles were also determined by the principal
12 investigator. For four agents, asbestos, wood dust, diesel exhaust, and formaldehyde, the
13 industries and occupations determined to have exposure were identified, and linking of
14 specific industries and occupations was based on literature review and consultation with local
15 industrial hygienists.
16 The relative odds were calculated for suspect occupations and industries classifying
17 individuals as ever employed for at least 1 year in an industry or occupation and defining the
18 reference group as those subjects never employed in that particular industry or occupation.
19 Multiple logistic regression models were used to control simultaneously for age, ethnicity,
20 and smoking status. For occupations with potential diesel exhaust exposure, the analysis
21 showed no excess risks for diesel engine mechanics and auto mechanics. Similarly, when
22 analyzed by exposure to specific agents, the odds ratio adjusted for age, smoking, and
23 ethnicity was not elevated for diesel exhaust fumes (OR = 0.6, 95% CI = 0.2, 1.6).
24 Elevated odds ratios were found for uranium miners (OR = 2.8, 95% CI = 1.0, 7.7),
25 underground miners (OR = 2.4, 95% CI = 1.2, 4.4), construction painters (OR = 2.4,
26 95% CI = 0.6, 9.6), and welders (OR = 4.3, 95% CI = 1.6, 11.0). No excess risks were
27 detected for the following industries: shipbuilding, petroleum refining, construction,
28 printing, blast furnace, and steel mills; or for the following occupations: construction
29 workers, painters, plumbers, paving equipment operators, roofers, engineers and firemen,
30 woodworkers, and shipyard workers. Females were excluded from detailed analysis because
31 none of the Hispanic female controls had been employed in high-risk jobs; among the
December 1994 8-28 DRAFT-DO NOT QUOTE OR CITE
-------
1 non-Hispanic white controls, employment in a high-risk job was recorded for at least five
2 controls for only two industries, construction and painting, for which the odds ratios were
3 not significantly elevated. Therefore, the analyses were presented for males only.
4 Among the many strengths of this study are its population-based design, high
5 participation rate, detailed smoking history, and the separate analysis done for the two ethnic
6 groups, southwestern Hispanic and non-Hispanic white males. The major limitations pertain
7 to the occupational exposure date. Job titles obtained from occupational histories were used
8 as proxy for exposure status, but these were not validated. Further, for nearly half the
9 cases, next of kin provided occupational histories. The authors acknowledge the above
10 sources of bias but state without substantiation that these biases would not strongly affect
11 their results. They also did not use a job exposure matrix to link occupations to exposures
12 and did not provide details on the method they used to classify individuals as diesel exhaust-
13 exposed based on reported occupations. The observed absence of an association for exposure
14 to asbestos, a well-established lung carcinogen, may be explained by the misclassification
15 errors in exposure status or by sample size constraints (not enough power). Likewise, the
16 association for diesel exhaust reported by only 7 cases and 17 controls also may have gone
17 undetected because of low power. In conclusion, there is insufficient evidence from this
18 study to confirm or refute an association between lung cancer and diesel exhaust exposure.
19
20 8.3.5 Garshick et al. (1987): A Case-Control Study of Lung Cancer and
21 Diesel Exhaust Exposure in Railroad Workers
22 An earlier pilot study of the mortality of railroad workers by the same investigators
23 (Schenker et al., 1984) found a moderately high risk of lung cancer among the workers who
24 were exposed to diesel exhaust as compared to those who were not. This study was designed
25 to evaluate the feasibility of conducting a large retrospective cohort study. On the basis of
26 these findings the investigators conducted a case-control study of lung cancer in the same
27 population. The population base for this case-control study of lung cancer and diesel exhaust
28 was approximately 650,000 active and retired male U.S. railroad workers with 10 years or
29 more of railroad service who were born in 1900 or later. The U.S. Railroad Retirement
30 Board (RRB), which operates the retirement system, is separate from the Social Security
31 System, and to qualify for the retirement or survivor benefits the workers had to acquire
December 1994 8-29 DRAFT-DO NOT QUOTE OR CITE
-------
1 10 years or more of service. Information on deaths that occurred between March 1, 1981,
2 and February 28, 1982, was obtained from the RRB. For 75% of the deceased population,
3 death certificates were obtained from the RRB, and, for the remaining 25%, they were
4 obtained from the appropriate state departments of health. Cause of death was coded
5 according to the eighth revision of the ICD. The cases were selected from deaths with
6 primary lung cancer, which was the underlying cause of death in most cases. Each case was
7 matched to two deceased controls whose dates of birth were within 2.5 years of the date of
8 birth of the case and whose dates of death were within 31 days of the date of death noted in
9 the case. Controls then were selected randomly from workers who did not have cancer noted
10 anywhere on their death certificates and who did not die of suicide or of accidental or
11 unknown causes.
12 Each subject's work history was determined from a yearly job report filed by his
13 employer with the RRB from 1959 until death or retirement. The year 1959 was chosen as
14 the effective start of diesel exhaust exposure for this study, since by this time 95% of the
15 locomotives in the United States were diesel powered. Investigators acknowledge that
16 because the transition to diesel-powered engines took place in the early 1950s, some workers
17 had additional exposure prior to 1959; however, if a worker had died or retired prior to
18 1959, he was considered unexposed. Exposure to diesel exhaust was considered to be
19 dichotomous for this study, which was assigned based on an industrial hygiene evaluation of
20 jobs and work areas. Selected jobs with and without regular diesel exhaust exposure were
21 identified by a review of job title and duties. Personal exposure was assessed in 39 job
22 categories representative of workers with and without diesel exhaust exposure. Those jobs
23 for which no personal sampling was done were considered exposed or unexposed on the basis
24 of similarities in job activities and work locations and by degree of contact with diesel
25 equipment. Asbestos exposure was categorized on the basis of jobs held in 1959 or on the
26 last job held if the subject retired before 1959. Asbestos exposure in railroads occurred
27 primarily during the steam engine era and was related mostly to the repair of locomotive
28 steam boilers that were insulated with asbestos. Smoking history information was obtained
29 from the next of kin.
30 Death certificates were obtained for approximately 87% of the 15,059 deaths reported
31 by the RRB from which 1,374 cases of lung cancer were identified. Fifty-five cases of lung
December 1994 8-30 DRAFT-DO NOT QUOTE OR CITE
-------
1 cancer were excluded from the study for either incomplete data (20) or refusal by two states
2 to use information on death certificates to contact the next of kin. Successful matching to at
3 least one control with work histories was achieved for 335 (96%) cases <64 years of age at
4 death and 921 (95%) cases >65 years of age at death. In both age groups, 90% of the cases
5 were matched with two controls. There were 2,385 controls in the study, 98% were
6 matched within ±31 days of the date of death, whereas the remaining 2% were matched
7 within 100 days. Deaths from diseases of the circulatory system predominated among
8 controls. Among the younger workers, approximately 60% had exposure to diesel exhaust,
9 whereas, among older workers, only 47% were exposed to diesel exhaust.
10 Analysis by a regression model, in which years of diesel exhaust exposure was the sum
11 total of the number of years in diesel-exposed jobs, used as a continuous exposure variable,
12 yielded an odds ratio of lung cancer to be 1.39 (95% CI = 1.05, 1.83) for over 20 years of
13 diesel exhaust exposure in the <64 years of age group. After adjustment for asbestos
14 exposure and lifetime smoking (pack-years), the odds ratio was 1.41 (95% CI = 1.06, 1.88).
15 Both crude odds ratio and asbestos exposure as well as lifetime smoking adjusted odds ratio
16 for the >65 years of age group were not significant. Increasing years of diesel exhaust
17 exposure categorized as >:20 diesel years and 5 to 19 diesel years with 0 to 4 years as the
18 referent group showed significantly increased risk in the <64 years of age group after
19 adjusting for asbestos exposure and pack-year category of smoking. For individuals who had
20 >20 years of diesel exhaust exposure, the odds ratio was 1.64 (95% CI = 1.18, 2.29),
21 whereas individuals who had 5 to 19 years of diesel exhaust exposure, the odds ratio was
22 1.02 (95% CI = 0.72, 1.45). In the >65 years of age group, only 3% of the workers were
23 exposed to diesel exhaust for more than 20 years. Relative odds for 5 to 19 years and
24 >20 years of diesel exposure were less than 1 (p > 0.01), after adjusting for smoking and
25 asbestos exposure.
26 Alternate models to explain post-asbestos exposure were tested. These were variables
27 for regular and intermittent exposure groups and an estimate of years of exposure based on
28 estimated years worked prior to 1959. No difference in results were seen. The interaction
29 between diesel exhaust exposure and the three pack-year categories ([1] <50, [2] >50, and
30 [3] missing pack-years) were explored. The cross-product terms were not significant.
31 A model was also tested that excluded recent diesel exhaust exposure occurring within the
December 1994 8-31 DRAFT-DO NOT QUOTE OR CITE
-------
1 5 years before death and gave an odds ratio of 1.43 (95% CI = 1.06, 1.94) adjusted for
2 cigarette smoking and asbestos exposure, for workers with 15 years of cumulative exposure.
3 For workers with 5 to 14 years of cumulative exposure, the relative odds were not
4 significant.
5 The many strengths of the study are consideration of confounding factors such as
6 asbestos exposure and smoking; classification of diesel exhaust exposures by job titles and
7 industrial hygiene sampling; exploration of interactions between smoking, asbestos exposure,
8 and diesel exhaust exposure; and good ascertainment (87%) of death certificates from the
9 15,059 deaths reported by the RRB.
10 The investigators also recognized and reported the following limitations:
11 overestimation of cigarette consumption by surrogate respondents which may have
12 exaggerated the contribution of smoking to lung cancer risk and use of the Interstate
13 Commerce Commission (ICC) job classification as a surrogate for exposure which may have
14 lead to misclassification of diesel exhaust exposure jobs with low intensity and intermittent
15 exposure, such as railroad police and bus drivers, as unexposed. These two limitations
16 would result in the underestimation of the lung cancer risk. This source of error could have
17 been avoided if diesel exhaust exposures were categorized by a specific dose range associated
18 with a job title that could have been classified as heavy, medium, low, and zero exposure
19 instead of a dichotomous variable. The use of death certificates to identify cases and controls
20 may have resulted in misclassification. Controls may have had undiagnosed primary lung
21 cancer, and lung cancer cases might have been secondary lesions misdiagnosed as primary
22 lung cancer. However, the investigators quote a third National Cancer Survey report in
23 which the death certificates for lung cancer were coded appropriately in 95% of the cases.
24 Lastly, as in all previous studies, there is a lack of data on the contribution of unknown
25 occupational or environmental exposures and passive smoking. In conclusion, this study,
26 compared with previous studies (on diesel exposure and lung cancer risk), provides the most
27 valid evidence that occupational diesel exhaust emission exposure increases the risk of lung
28 cancer.
29
30
December 1994 g_32 DRAFT-DO NOT QUOTE OR CITE
-------
1 8.3.6 Benhamou et al. (1988): Occupational Risk Factors of Lung Cancer
2 in a French Case-Control Study
3 This is a case-control study of 1,625 histologically confirmed cases of lung cancer and
4 3,091 matched controls, conducted in France between 1976 and 1980. This study was part
5 of an international study to investigate the role of smoking and lung cancer. Each case was
6 matched with one or two controls whose diseases were not related to tobacco use, sex, age at
7 diagnosis (±5 years), hospital of admission, or interviewer. Information was obtained from
8 both cases and controls on place of residence since birth, educational level, smoking, and
9 drinking habits. A complete lifetime occupational history was obtained by asking participants
10 to give their occupations from the most recent to the first. Women were excluded because
11 most of them had listed no occupation. Men who smoked cigars and pipes were excluded
12 because there were very few in this category. Thus, the study was restricted to nonsmokers
13 and cigarette smokers. Cigarette smoking exposure was defined by age at the first cigarette
14 (nonsmokers, ^20 years, or >20 years), daily consumption of cigarettes (nonsmokers,
15 <20 cigarettes a day, and >20 cigarettes a day), and duration of cigarette smoking
16 (nonsmokers, <35 years, and >35 years). The data on occupations were coded by a panel
17 of experts according to their own chemical or physical exposure determinations. Occupations
18 were recorded blindly using the International Standard Classification of Occupations. Data
19 on 1,260 cases and 2,084 controls were available for analysis. The remaining 365 cases and
20 1,007 controls were excluded because they did not satisfy the required smoking status
21 criteria.
22 A matched logistic regression analysis was performed to estimate the effect of each
23 occupational exposure after adjusting for cigarette status. Matched relative risk (RR) ratios
24 were calculated for each occupation with the baseline category, which consisted of patients
25 who had never been engaged in that particular occupation. The matched relative risk ratios
26 adjusted for cigarette smoking for the major groups of occupations showed that the risks
27 were significantly higher for production and related workers, transport equipment operators,
28 and laborers (RR = 1.24, 95% CI = 1.04, 1.47). On further analysis of this group, for
29 occupations with potential diesel emission exposure, significant excess risks were found for
30 motor vehicle drivers (RR = 1.42, 95% CI = 1.07, 1.89) and transport equipment operators
31 (RR = 1.35, 95% CI = 1.05, 1.75). No interaction with smoking status was found in any
December 1994 3.33 DRAFT-DO NOT QUOTE OR CITE
-------
1 of the occupations. The only other significant excess was observed for mines and quarrymen
2 (RR = 2.14, 95% CI = 1.07, 4.31). None of the significant associations showed a dose-
3 response relationship with duration of exposure.
4 This study was designed primarily to investigate the relationship between smoking (not
5 occupations and environmental exposures) and lung cancer. Although an attempt was made
6 to obtain complete occupational histories, the authors did not clarify if in the logistic
7 regression analysis, they used the subjects first occupation, predominant occupation, last
8 occupation, or ever worked in that occupation as the risk factor of interest. The most
9 important limitation of this study is that the occupations were not coded into exposures for
10 different chemical and physical agents, thus precluding the calculation of relative risks for
11 diesel exposure. Using occupations as surrogate measures of diesel exposure, an excess
12 significant risk was obtained for motor vehicle drivers and transport equipment operators, but
13 not for motor mechanics. However, it is not known if subjects in these occupations worked
14 with diesel engines or nondiesel engines.
15
16 8.3.7 Hayes et al. (1989): Lung Cancer in Motor Exhaust-Related
17 Occupations
18 This study reports the findings from an analysis of pooled data from three lung cancer
19 case-control studies that examine in detail the association between employment in motor
20 exhaust-related (MER) occupations and lung cancer risk adjusted for confounding by smoking
21 and other risk factors. The three studies were carried out by the National Cancer Institute in
22 Florida (1976 to 1979), New Jersey (1980 to 1981), and Louisiana (1979 to 1983). These
23 three studies were selected because the combined group would provide a sufficient sample to
24 detect a risk of lung cancer in excess of 50% among workers in MER occupations. The
25 analyses were restricted to males who had given occupational history. The Florida study was
26 hospital-based with cases ascertained through death certificates. Controls were randomly
27 selected from hospital records and death certificates, excluding psychiatric diseases matched
28 by age and county. The New Jersey study was population-based with cases ascertained
29 through hospital records, cancer registry, and death certificates. Controls were selected from
30 among the pool of New Jersey licensed drivers and death certificates. The Louisiana study
31 was hospital-based (it is not specified how the cases were ascertained), and controls were
December 1994 8-34 DRAFT-DO NOT QUOTE OR CITE
-------
1 randomly selected from hospital patients, excluding those with lung diseases and tobacco-
2 related cancers.
3 A total of 2,291 cases of male lung cancers and 2,570 controls were eligible, and the
4 data on occupations were collected by next-of-kin interviews for all jobs held for 6 mo or
5 more, including the industry, occupation, and number of years employed. The proportion of
6 next-of-kin interviews varied by site between 50% in Louisiana to 85% in Florida. The
7 coding schemes were reviewed to identify MER occupations, which included truck drivers
8 and heavy equipment operators (cranes, bulldozers, and graders); bus drivers, taxi drivers,
9 chauffeurs, and other motor vehicle drivers; and automobile and truck mechanics. Truck
10 drivers were classified as routemen and delivery men and other truck drivers. All jobs were
11 also classified with respect to potential exposure to known and suspected lung carcinogens.
12 Odds ratios were calculated by the maximum likelihood method adjusting for age by birth
13 year, usual amount smoked, and study area. Logistic regression models were used to
14 examine the interrelationship of multiple variables.
15 A statistically significant excess risk was detected for employment of 10 years or more
16 for all MER occupations (except truck drivers) adjusted for birth cohort, usual daily cigarette
17 use, and study area. The odds ratio for lung cancer using data gathered by direct interviews
18 was 1.4 (95% CI = 1.1, 2.0), allowing for multiple MER employment and 2.0 (95%
19 CI = 1.3, 3.0) excluding individuals with multiple MER employment. Odds ratios for all
20 MER employment, except truck drivers who were employed for less than 10 years, were
21 1.3 (95% CI = 1.0, 1.7) and 1.3 (95% CI = 0.9, 1.8) for including and excluding multiple
22 MER employment, respectively. Odds ratios were then derived for specific MER
23 occupations, and, to avoid the confounding effects of multiple MER job classifications,
24 analyses were also done excluding subjects with multiple MER job exposures. Truck drivers
25 employed for more than 10 years had an odds ratio of 1.5 (95% CI = 1.1, 1.9). A similar
26 figure was obtained excluding subjects with multiple MER employment. An excess risk was
27 not detected for truck drivers employed less than 10 years. The only other job category that
28 showed a statistically significant excess for lung cancer was the one that included taxi drivers
29 and chauffeurs who worked multiple MER jobs for less than 10 years (OR = 2.5, 95%
30 CI = 1.4, 4.8). For the same category the risk for individuals working in that job for more
31 than 10 years was 1.2 (95% CI = 0.5, 2.6). A statistical significant positive trend
December 1994 3.35 DRAFT-DO NOT QUOTE OR CITE
-------
1 (p < 0.05) with increasing employment of <2 years, 2 to 9 years, 10 to 19 years, and
2 20+ years was observed for truck drivers but not for other MER occupations. A statistically
3 nonsignificant excess risk was also observed for heavy equipment operators, bus drivers, taxi
4 drivers and chauffeurs, and mechanics employed for 10 years or more. All of the above-
5 mentioned odds ratios were derived adjusted for birth cohort, usual daily cigarette use, and
6 state of residence. Exposure to other occupational suspect lung carcinogens did not account
7 for the excess risks detected.
8 Results of this large study provide evidence that workers in MER jobs are at an excess
9 risk of lung cancer that is not explained by their smoking habits or exposures to other lung
10 cancers. Because no information on type of engine had been collected, it was not possible to
11 determine if the excess risk was due to exposure to diesel exhaust or gasoline exhaust or the
12 mixture of the two. Among its limitations are possible bias due to misclassification of jobs
13 reported by the large proportion of next-of-kin interviews and the problems in classifying
14 individuals into uniform occupational groups based on the pooled data in the three studies
15 that used different occupational classification schemes.
16
17 8.3.8 Steenland et al. (1990): A Case-Control Study of Lung Cancer and
18 Truck Driving in the Teamsters Union
19 Steenland et al. conducted a case-control study of lung cancer deaths in the Teamsters
20 Union to determine the risk of lung cancer among different occupations. Death certificates
21 were obtained from the Teamsters Union files in the central states for 10,485 (98%) male
22 decedents who had filed claims for pension benefits and who had died in 1982 and 1983.
23 Individuals were required to have 20 years tenure in the union to be eligible to claim
24 benefits. Cases comprised all deaths (n = 1,288) from lung cancer, coded as ICD 162 or
25 163 for underlying or contributory cause on the death certificate. The 1,452 controls
26 comprised every sixth death from the entire file excluding deaths from lung cancer, bladder
27 cancer, and motor vehicle accidents. Detailed information on work history and potential
28 confounders such as smoking, diet, and asbestos exposure was obtained by questionnaire.
29 Seventy-six percent of the interviews were provided by spouses and the remainder by some
30 other next of kin. The response rate was 82% for cases and 80% for controls. Using these
31 interview data and the 1980 census occupation and industry codes, subjects were classified
December 1994 8-36 DRAFT-DO NOT QUOTE OR CITE
-------
1 either as nonexposed, or as having held other jobs with potential diesel exhaust exposure.
2 Data on job categories were missing for 12% of the study subjects. A second work history
3 file was also created based on the Teamsters Union pension application that lists occupation,
4 employer, and dates of employment. A three-digit U.S. census code for occupation and
5 industry was assigned to each job for each individual. This Teamsters Union work history
6 file did not have information on whether men drove diesel or gasoline trucks, and the four
7 principal occupations were long-haul drivers, short-haul or city drivers, truck mechanics, and
8 dock workers. Subjects were assigned the job category in which they had worked the
9 longest.
10 The case-control analysis was done using unconditional logistic regression. Separate
11 analyses were conducted for work histories from the Teamsters Union pension file and from
12 next-of-kin interviews. Covariate data were obtained from next-of-kin interviews. Analyses
13 were also performed for two time periods: employment after 1959 and employment after
14 1964. These two cut-off years reflect years of presumed dieselization; 1960 for most
15 trucking companies and 1965 for independent driver and nontrucking firms. Data for
16 analysis could be obtained for 994 cases and 1,085 controls using Teamsters Union work
17 history and for 872 cases and 957 controls, using next-of-kin work history. When exposure
18 was considered as a dichotomous variable, for both Teamsters Union and next-of-kin work
19 history, no single job category had an elevated risk. From the next-of-kin data, diesel truck
20 drivers had an odds ratio of 1.42 (95% CI = 0.74, 2.47) and diesel truck mechanics had an
21 odds ratio of 1.35 (95% CI = 0.74, 2.47). Odds ratios by duration of employment as a
22 categorical variable were then estimated. For the Teamsters Union work history data and
23 when only employment after 1959 was considered, both long haul (p < 0.04) and short haul
24 drivers (not significant) showed an increase in risk with increased years of exposure. The
25 length of employment categories for which the trends were analyzed were 1 to 11 years,
26 12 to 17 years, and 18 years or more. Using 1964 as the cut-off date, long haul drivers
27 continued to show a significant positive trend (p = 0.04) with an odds ratio of
28 1.64 (95% CI = 1.05, 2.57) for those who worked for 13+ years, the highest category.
29 Short haul drivers, however, did not show a positive trend when 1964 was used as the cut-
30 off date. Similar trend analysis was done for most next-of-kin data. A marginal increase in
31 risk with increasing duration of employment as a truck driver (p = 0.12) was observed. For
December 1994 8-37 DRAFT-DO NOT QUOTE OR CITE
-------
1 truck drivers who primarily drove diesel trucks for 35 years or longer, the odds ratio for
2 lung cancer was 1.89 (95% CI = 1.04, 3.42). The odds ratio for gasoline truck drivers was
3 1.34 (95% CI = 0.81, 2.22) and for truck mechanics was 1.09 (95% CI = 0.44, 2.66).
4 No significant interactions between age and diesel exhaust exposure or smoking and diesel
5 exhaust exposure were observed. All the odds ratios were adjusted for age, smoking, and
6 asbestos in addition to various exposure categories.
7 The authors acknowledge several limitations of this study which include possible
8 misclassifications of exposure and smoking habits, as information was provided by next of
9 kin; lack of sufficient latency to observe lung cancer excess; and a small nonexposed group
10 (n = 120). Also, concordance between Teamsters Union and next-of-kin job categories
11 could not be easily evaluated because job categories were defined differently in each data set.
12 No data were available on levels of diesel exposure for the different job categories. Given
13 these limitations, the positive findings of this study are probably underestimated.
14 Table 8-2 summarizes the lung cancer case control studies.
15
16
17 8.4 CASE-CONTROL STUDIES OF BLADDER CANCER
18 8.4.1 Howe et al. (1980): Tobacco Use, Occupation, Coffee, Various
19 Nutrients, and Bladder Cancer
20 This is a Canadian population-based case-control study conducted in the provinces of
21 British Columbia, Newfoundland, and Nova Scotia. These areas were selected because they
22 had cancer registries and were believed not to have concentrations of high-risk industries.
23 All patients with newly diagnosed bladder cancer occurring in the three provinces between
24 April 1974 and June 1976 were identified, and 77% of them were interviewed at home.
25 A total of 480 male and 152 female case-control pairs were available for analysis. For each
26 case one neighborhood control, matched by age (±5 years) and sex, was also interviewed at
27 home to obtain data on smoking, occupation, dietary sources of nitrites and nitrates which
28 convert to nitrosamines (nonpublic water supply and preserved meat products), and beverage
29 consumption, including a detailed history of coffee consumption. A detailed smoking history
30 was obtained. The occupational history included a chronological account of all jobs and the
December 1994 8-38 DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 8-2. EPIDEMIOLOGIC STUDIES OF THE HEALTH EFFECTS OF EXPOSURE TO DIESEL EXHAUST:
CASE-CONTROL STUDIES OF LUNG CANCER
Authors
Population Studied
Diesel Exhaust Exposure
Assessment
Results
Limitations
Williams 7,518 (3,539 males and
et al. (1977) 3,979 females) incident invasive
cancers from the Third National
Cancer Survey
Lung cancer cases:
32 in males
28 in females
Combined other cancer sites
were used as controls.
Main lifetime, recent,
and other employment
information obtained at
the time of survey
1970 Census Coding
Scheme for Employment
was used to code the
occupations by one of
the authors.
SNS elevated relative
odds were observed
among occupations of
trucking, railroading,
and mining.
Exposure estimation based on self-report that was
not validated
47% Nonresponse
Control group consisted of other cancers
probably diluting the risk estimation.
Small numbers in cause- specific cancers and
individual occupations
9°
Hall and 502 Histologically confirmed
Wynder lung cancers
(1984) Cases diagnosed 12 mo prior to
interviews
502 Matched hospital controls
without tobacco related diseases,
matched for age, sex, race, and
geographical area
Population from 18 hospitals in
controls
Based on previous
Industrial Hygiene
Standards for a
particular occupation,
usual lifetime occupation
coded as "probably high
exposure" and "no
exposure"
N1OSH standards used
to classify exposures:
High
Moderate
Low
SNS excess risk after
adjustment for smoking
for lung cancer:
RR = 1.4 (1st criteria)
and
RR = 1.7(NIOSH
criteria)
Complete lifetime employment history not
available
Self-reported occupation history not validated
No analysis by dose, latency, or duration of
exposure
No information on nonoccupational diesel
exposure
-------
o
CD
I
TABLE 8-2 (cont'd). EPIDEMIOLOGIC STUDIES OF THE HEALTH EFFECTS OF EXPOSURE TO DIESEL
EXHAUST: CASE-CONTROL STUDIES OF LUNG CANCER
Authors
Population Studied
Diesel Exhaust Exposure
Assessment
Results
Limitations
oo
-^
o
Damber and 589 Lung cancer cases who had
Larsson died prior to 1979 reported to
(1987) Swedish registery between 1972
and 1977
582 Matched dead controls (sex,
age, year of death, municipality)
drawn from National Registry
of Cause of Death
453 Matched living controls
(sex, year of birth, municipality)
drawn from National
Population Registry
Occupations held for at
least 1 year or more
Using a 5-digit code the
occupations were
classified according to
Nordic Classification of
Occupations
SS OR = 2.7
of employment)
SS OR = 9.8 (>20 years
of employment)
Adjustment for smoking
was done.
SNSOR = 1.2 for
professwional drivers (>20
years of employment) with
dead controls
SNSOR = 1.1 >20 years
of employment) with
living controls.
year Uncertain diesel exhaust exposure
No validation of exposure done
Underground miners data not adjusted for other
confounders such as radon, etc.
O
O
O
I
3
O
H
W
Lerchen 506 Lung cancer cases from
et al. (1987) New Mexico Tumor registry
(333 males and 173 females)
Aged 25-84 years
Diagnosed between January 1,
1980, and December 31, 1982
771 (499 males and 272 females)
frequency matched with cases,
selected from telephone directory
Lifetime occupational
history and selfreported
exposure history was
obtained.
Coded according to
Standard Industrial
Classification Scheme
No excess of relative
odds was observed for
diesel exhaust exposure.
Exposure based on occupational history and self
report, which was not validated
50% Occupational history provided by next of
kin
Absence of lung cancer association with asbestos
suggests misclassification of exposure.
-------
I
TABLE 8-2 (cont'd). EPEDEMIOLOGIC STUDIES OF THE HEALTH EFFECTS OF EXPOSURE TO DIESEL
EXHAUST: CASE-CONTROL STUDIES OF LUNG CANCER
Authors
Population Studied
Diesel Exhaust Exposure
Assessment
Results
Limitations
Garshick 1,319 Lung cancer cases who
et al. (1987) died between March 1, 1981,
and February 28, 1982
2,385 Matched controls (two
each, age and date of death)
Both cases and controls drawn
from railroad worker cohort
who had worked for 10 or
more years
Personal exposure assessed SS OR = 1.41 (^64 year Probable misclassification of diesel exhaust
for 39 job categories. age group exposure jobs
This was corrected with
job titles to dichotomize
the exposure into:
Exposed
Not exposed
SS OR = 1.64 (<64 year
age group) for >20 years
diesel exhaust exposure
group when compared to
0- to 4-year exposure
group
All ORs adjusted for life
time smoking and asbestos
exposure.
Years of exposure used as surrogate for dose
13% of death certificates not ascertained
Overestimation of smoking history
oo
o
3
Benhamou 1,260 Histologically confirmed
et al. (1988) lung cancer cases
2,084 Non-tobacco-related
disease matched controls
(sex, age at diagnosis,
hospital admission, and
interviewer)
Occurring between 1976 and
1980 in France
Based on exposures
determined by panel of
experts
The occupations were
recorded blindly using
International Standard
Classification of
Occupations as chemical
or physical exposures.
Significant excess risks
were found in motor
vehicle drivers
(RR = 1.42) and
transport equipment
operators (RR = 1.35)
(smoking adjusted).
Exposure based on occupational histories not
validated
Exposures classified as chemical and physical
exposure, not specific to diesel exhaust
O
d
o
o
I-H
H
w
Hayes et al. Pooled data from three different
(1989) studies consisting of 2,291 male
lung cancer cases
2,570 Controls
Occupational information
from next of kin for all
jobs held
Jobs classified with respect
to potential exposure to
known and suspected
pulmonary carcinogens
SS OR = 1.5 for truck
drivers (> 10 years of
employment)
SS positive trend with
increasing employment as
truck driver
Exposure data based on job description given by
next of kin, which was not validated
Could have been mixed exposure to both diesel
and gasoline exhausts
Job description could have lead to
misclassification
-------
o
n
o
n
I
TABLE 8-2 (cont'd). EPIDEMIOLOGIC STUDIES OF THE HEALTH EFFECTS OF EXPOSURE TO DIESEL
EXHAUST: CASE-CONTROL STUDIES OF LUNG CANCER
Authors
Population Studied
Diesel Exhaust Exposure
Assessment
Results
Limitations
oo
ife
Steenland 1,058 Male lung cancer deaths
et al. (1990) between 1982 and 1983
1,160 Every sixth death from
entire mortality file sorted by
social security number
(excluding lung cancer,
bladder cancer, and motor
vehicle accidents)
Cases and controls were from
Central State Teamsters who
had filed claims (requiring
20-year tenure).
Longest job held: diesel As 1964 cutoff point:
truck driver, gasoline
truck driver, both types SS OR = 1.64 for long
of trucks, truck haul drivers with
mechanic, and dock 13+ years of employment
workers
Positive trend test for long
haul drivers (p = 0.04)
SSOR = 1.89 for diesel
truck drivers of 35 + years
of employment
Exposure based on job titles not validated
Possible misclassification of exposure and
smoking, based on nextofkin information
Lack of sufficient latency
Abbreviations:
OR = Odds ratio.
RR = Relative risk.
SNS = Statistically nonsignificant.
SS = Statistically significant.
-------
1 number of years and months during which the respondent had worked in each job, experience
2 in industries that were suspected a priori to increase the risk of bladder cancer, and exposure
3 to any jobs that involved exposure to dust and fumes at the workplace. Relative risk
4 estimates were computed using the linear logistic model applied to individually matched case-
5 control pairs.
6 A base-line comparison of cases and controls showed that, whereas male patients were
7 similar to controls on income and education, there was an excess of female cases with low
8 family incomes and low levels of educational attainment. For both sexes the mean ages for
9 cases and controls did not differ, and the times required for the interview were similar.
10 An analysis by the a priori suspect industries showed elevated risks for a number of
11 industries for males. These included the chemical (RR = 7.5, 95% CI = 1.7, 67.6),
12 rubber (RR = 5.0, 95% CI = 0.6, 236.5), petroleum (RR = 5.3, 95% CI = 1.5, 28.6),
13 medicine (RR = 2.6, 95% CI = 0.9, 9.3), and spray painting (RR = 1.8, 95% CI = 0.7,
14 4.6) industries. The excess risks were statistically significant only for the petroleum and
15 chemical industries. The estimates did not change when the analysis was done separately for
16 subjects who reported only one exposure and for those who reported exposure to more than
17 one suspect industry. The estimates also remained unchanged after controlling for smoking.
18 Too few females reported working in the a priori suspect industries to make any meaningful
19 contribution to the analysis. Among males, statistically nonsignificant excess risks were
20 observed for tanning, electric cable, photographic, commercial paint, tailoring, medicine,
21 food processing, and agricultural industries. The analysis by exposure to dust and fumes in
22 occupations other than those in the a priori suspect list detected the relative risks for diesel
23 and traffic fumes (RR = 2.8, 95% CI = 0.8, 11.8). Statistically significant excess risks
24 were observed for railroad workers (RR = 9.0, 95% CI = 1.2, 394.5) and welders
25 (RR = 2.8, 95% CI = 1.1, 8.8). For occupations other than those on the a priori list for
26 males and females, statistically significant excesses were detected for metal machinists
27 (RR = 2.7, 95% CI = 1.1, 7.6), metal recorders (RR = 2.6, 95% CI = 1.0, 7.3), and
28 nursery men (RR = 5.5, 95% CI = 1.2, 51.1). Statistically nonsignificant excesses were
29 also detected for exposure to two chemicals: benzidine and its salts, RR = 1.3, and
30 to-chloromethyl ether, RR = 5.0. A detailed analysis was done for cigarette smoking,
31 which demonstrated statistically significant increasing bladder cancer risk with increasing
December 1994 8-43 DRAFT-DO NOT QUOTE OR CITE
-------
1 duration of smoking, total lifetime consumption of packs of cigarettes, and average frequency
2 of cigarettes per day. In males the highest significant risk was observed for latency of less
3 than 35 years; after that time the risk reduced slightly with increasing latency. In females
4 the highest significant risk was for more than 35 years of latency. Risks were elevated for
5 males consuming all types of coffee and for females consuming instant coffee. Hair dye
6 usage in females and phenacetin usage in males and females carried no risk. Significant risks
7 for use of artificial sweeteners and use of nonpublic water supplies (nitrates and nitrites)
8 were found among males only.
9 This study was mainly designed to evaluate the various risk factors for bladder cancer
10 such as smoking, coffee consumption, nitrates and nitrites in diet, etc. The major limitation
11 of this study, as the authors noted, was that the three selected provinces did not have high
12 concentrations of industries suspected to be linked to bladder cancer. An excess risk was,
13 however, detected for railroad workers and for those in the "exposed to diesel and traffic
14 fumes category." Risks for those exposed to "diesel fumes only" were not available nor do
15 we know the exact job title of the railroad workers and the type of engines they were
16 operating. The authors also did not detail the method by which they coded the information
17 given by respondents in response to questions on exposure to dust and fumes into the various
18 categories they used in the analysis. These analyses were done for subjects who reported
19 having "ever been exposed" versus "never been exposed" to these fumes, and, although
20 detailed chronological work histories were obtained, no attempt was made to develop a
21 lifetime cumulative exposure index to diesel fumes. In multiple logistic regression models,
22 the authors used the a priori high-risk occupations; hence, nothing can be concluded about
23 exposure to diesel exhaust for occupations that were not part of that list. The authors
24 provided no explanation on possible selection bias as only 77% of the eligible population was
25 included in the study.
26
27 8.4.2 Wynder et al. (1985): A Case-Control Study of Diesel Exhaust
28 Exposure and Bladder Cancer
29 A case-control study of diesel exhaust exposure and bladder cancer risk was conducted
30 by Wynder et al. (1985). Cases and controls were obtained from 18 hospitals located in six
31 U.S. cities between January 1981 and May 1983. Cases were individuals with histologically
December 1994 8-44 DRAFT-DO NOT QUOTE OR CITE
-------
1 confirmed primary cancer of the bladder, diagnosed within 12 mo prior to the interview.
2 Controls were individuals with nontobacco-related diseases who were matched to the cases by
3 age (within 8 years), race, year of interview, and hospital of admission. Women were
4 excluded from the study since the focus was on male-dominated occupations. A structured
5 questionnaire was administered in the hospital to cases and controls to elicit information on
6 usual occupation, smoking history, alcohol and coffee consumption, as well as other
7 demographic factors.
8 Two methods were used to define occupational exposure to diesel exhaust. First,
9 occupational titles defined by the industrial hygiene standards as probable high exposure were
10 classified as exposed or not exposed to diesel exhaust. The probable high exposure category
11 consisted of bus and truck drivers, heavy equipment operators and repairmen, railroad
12 workers, and warehousemen. In the second method, guidelines set by NIOSH were used to
13 classify occupations based on exposure to diesel exhaust. In this method, the estimated
14 proportion of exposed workers was computed for each occupational category by using the
15 NIOSH estimates of the exposed population as the numerator and the estimates of individuals
16 employed in each occupational category from the 1970 census as the denominator.
17 Occupations estimated to have at least 20% of their employees exposed to diesel exhaust
18 were defined as "high exposure," those with 10 to 19% of their employees exposed as
19 "moderate exposure," and those with less than 10% of their empolyees exposed as "low
20 exposure." The odds ratio was used as a measure of association to assess the relationship
21 between bladder cancer and diesel exhaust exposure. The overall participation among those
22 eligible and available for interview was 75 and 72% in cases and controls, respectively.
23 A total of 194 bladder cancer cases and 582 controls were examined, and the two
24 groups were found to be comparable by age and education. Except for railroad workers who
25 had relative odds of 2.0 based on two cases and three controls (95% CI = 0.34, 11.61), the
26 relative odds were less than one for other diesel exhaust exposure occupations such as bus
27 and truck drivers, warehousemen, material handlers, and heavy equipment workers. When
28 the risk was examined using the NIOSH criteria for high, moderate, and low exposure,
29 relative odds were 1.68 and 0.16 for high and moderate, respectively, with low as the'
30 referent group; neither was statistically significant. Cases and controls were compared by
31 smoking status. Cases were more likely to be current cigarette smokers than controls.
December 1994 8-45 DRAFT-DO NOT QUOTE OR CITE
-------
1 Current smokers of 1 to 20 cigarettes/day had relative odds of 3.64 (95% CI = 2.04, 6.49),
2 current smokers of 21+ cigarettes/day had relative odds of 3.51 (95% CI = 2.00, 6.19),
3 while exsmokers had relative odds of 1.72 (95% CI = 1.01, 2.92). After controlling for
4 smoking, there was no significant increase in the risk of bladder cancer for occupations with
5 diesel exhaust exposure compared to occupations without diesel exhaust exposure.
6 A synergistic effect between the two also was not detected.
7 This study has two major methodologic limitations, both pertaining to exposure
8 classification. First, the use of "usual" occupation may have lead to misclassification of
9 those individuals who had held a previous job with diesel exhaust exposure that was not their
10 usual occupation; this may have resulted in reduced power to detect weak associations.
11 Second, since there was no information on amount or duration of diesel exhaust exposure, no
12 analysis of dose-response relationships could be done. Also, no information was available on
13 other confounding risk factors of bladder cancer such as urinary retention, amphetamine
14 abuse, and smoking within the confined space of a truck cab, all of which are life-style
15 factors specific to the truck driving occupation.
16
17 8.4.3 Hoar and Hoover (1985): Truck Driving and Bladder Cancer
18 Mortality in Rural New England
19 This study investigated the relationship between the occupation of truck driving and
20 bladder cancer mortality in a case-control study in New Hampshire and Vermont. Cases
21 included all white residents of New Hampshire and Vermont who died from bladder cancer
22 (eighth revision of the ICD) between 1975 and 1979. Death certificates were provided by
23 the vital records and health statistics office of the two states, and the next of kin were traced
24 and interviewed in person. Two types of controls were selected for each case. One control
25 was randomly selected from all other deaths, excluding suicides, and matched on state, sex,
26 race, age (±2 years), and year of death. The second control was selected with the additional
27 matching criteria of county of residence. Completed interviews were obtained from 325 (out
28 of 410) next of kin for cases and 673 (out of 923) controls. Information on demographic
29 characteristics, lifetime occupational and residential histories, tobacco use, diet, and medical
30 history were obtained on each subject. The odds ratio was calculated to ascertain a measure
31 of association between truck driving and bladder cancer. Because separate analyses of the
December 1994 8-46 DRAFT-DO NOT QUOTE OR CITE
-------
1 two control series gave similar results, the two control series were combined. Also, because
2 matched analyses yielded results similar to those provided by the unmatched analyses, results
3 of the unmatched analyses were presented.
4 Sixteen percent (35) of the cases and 12% (53) of the controls had been employed as
5 truck drivers yielding an odds ratio of 1.5 (95% CI = 0.9, 2.6) after adjustment for county
6 of residence and age at death. For New Hampshire, the odds ratio was
7 1.3 (95% CI = 0.7, 2.3), and for Vermont the odds ratio was 1.7 (95% CI = 0.8, 3.4).
8 For a large number of subjects, the next of kin were unable to give the durations of truck
9 driving, and there was an inconsistent positive association with years of truck driving. Crude
10 relative odds were not altered after adjustment for coffee drinking, cigarette smoking, and
11 education as a surrogate for social class. Little variation in risks was seen when the data
12 were analyzed by the industry in which the men had driven trucks. No relationship was seen
13 between age at which employment as a truck driver started and occurrence of bladder cancer.
14 Analysis by duration of employment as a truck driver and bladder cancer showed a positive
15 trend of increasing relative odds with increasing duration of employment. The trend test was
16 statisically significant (p = 0.006). The odds ratio was statistically significant for the 5- to
17 9-years employment category only (OR = 2.9, 95% CI = 1.2, 6.7). Similarly, analysis by
18 calendar year first employed showed a statistically significant odds ratio for 1930 to 1949
19 (OR = 2.6, 95% CI = 1.3, 5.1), whereas relative odds were not significant if they were
20 employed prior to 1929 or after 1950.
21 The effects of reported diesel exhaust exposure from fuel or engines in truck driving or
22 other occupations were then analyzed. An odds ratio of 1.8 (95% CI = 0.5, 7.0) was
23 derived for those who were exposed to diesel exhaust during their truck driving jobs as
24 compared to an odds ratio of 1.5 (95% CI = 0.8, 2.7) for those not reporting diesel exhaust
25 exposure. Analysis by duration of exposure (0, 1 to 19 years, 20 to 29 years, 30 to
26 39 years, and 40+ years) to diesel fuel or engines in other occupations, which were
27 "self-reported" by participants, showed a statistically significant positive trend (p = 0.024)
28 for bladder cancer, although none of the individual odds ratios in these duration categories
29 were statistically significant.
30 This study investigated an association between truck driving and bladder cancer in an
31 attempt to understand the reasons for the high rates of bladder cancer in rural areas of
December 1994 8-47 DRAFT-DO NOT QUOTE OR CITE
-------
1 New Hampshire and Vermont. Although an elevated odds ratio for bladder cancer (not
2 statistically significant) was observed for reported truck-driving occupations, there was
3 insufficient evidence to conclude that the excess risk of bladder cancer was due to exposure
4 to diesel emissions. This is because the excess bladder cancer risk was observed for all truck
5 drivers irrespective of their exposure to diesel emissions. Also, no information was provided
6 on the confounding effects of other aspects of the road environment such as urinary
7 retention, amphetamine abuse, and concentrated cigarette smoke within the truck cab. Other
8 limitations of this study include the use of next of kin for occupational histories who may
9 either under- or overreport occupations and the use of death certificates with their inherent
10 problems of misclassification.
11
12 8.4.4 Steenland et al. (1987): A Case-Control Study of Bladder Cancer
13 Using City Directories as a Source of Occupational Data
14 The primary objective of the study was to test the usefulness of city directories as a
15 source of occupational data in epidemiologic studies of illness and occupational exposure.
16 Commercial city directories provide data on occupations and employers and are compiled
17 from a door-to-door yearly census of all residents 18 years old and older. The directories
18 are available in most medium-size cities in the United States. A unique feature of city
19 directory data is that they identify specific employers, and as the authors suggest, this
20 information may be better than death certificates for rapid, inexpensive, record-based,
21 epidemiologic studies.
22 A case-control study was conducted of male bladder cancer deaths in Hamilton County
23 (including Cincinnati), OH. This county was selected because it is in an industrialized area
24 with high bladder cancer rates and also because city directories cover approximately 85% of
25 the people living in the county. A computerized list of all male bladder cancer deaths
26 (n = 731) and all other male deaths (n = 95,057), with the exclusion of deaths from urinary
27 tract tumors and pneumonia, that occurred between 1960 and 1982 was obtained from the
28 Ohio Department of Vital Statistics. Death certificates had been coded by a nosologist
29 according to the ICD code in use at the time of death. A pool of six controls was created for
30 each case matched on sex, residence in Hamilton County at time of death, year of death, age
31 at death (±5 years), and race.
December 1994 8-48 DRAFT-DO NOT QUOTE OR CITE
-------
1 Two types of analysis were performed, one based on city directory data and the other
2 based on death certificate data. In the former, cases and controls were restricted to
3 individuals who had at least one yearly directory listing with some occupational data. The
4 first two controls from the pool of six who met the requirements were selected. This
5 analysis involved 648 cases (627 cases had 2 controls and 21 cases had only 1 control) and
6 1,275 controls.
7 The death certificate analysis involved all 731 cancer deaths, with two controls per
8 case. In most cases, the same two controls were used in this analysis too. The usual
9 lifetime occupation and industry on the death certificate was abstracted from them. Data on
10 occupation and industry were coded with a three-digit U.S. census code using the method
11 adopted by the U.S. Bureau of the Census. Five of the occupational data were recorded for
12 occupation and industry by a second coder, with a high degree of reproducibility. Odds
13 ratios were calculated for bladder cancer using a Mantel-Haenszel procedure.
14 The city directory data identified four employers significantly associated with bladder
15 cancer deaths; only one of them was identified by the death certificate data that provided
16 only lifetime type of industry rather than the name of a specific employer. The industries
17 represented by the four employers were a chemical plant, printing company, valve company,
18 and machinery plant. The city directory data analysis demonstrated significant positive
19 associations for quite a few occupations. The occupations that had at least 10 cases or more
20 were engineers (OR = 3.00, p = 0.01), carpenters (OR = 2.36, p < 0.01), tailors
21 (OR = 2.56, p < 0.01), and furnace operators (OR = 2.5, p = 0.03). Relative odds were
22 increased significantly with increased duration of employment (^20 years) for truck drivers
23 (OR = 12, p = 0.01) and furnace operators (based on four cases and no controls,
24 p = 0.05). For occupations in which subjects had ever been employed, a significant
25 increase in the relative odds with increased duration of employment was observed for the
26 railroad industry (>20 years of employment, OR = 2.21, p < 0.05). Both truck driving
27 and railroad industry occupations involve diesel emission exposures.
28 The analysis of death certificate data yielded associations in the same direction for most
29 of the occupations. A check of the validity of city directory data indicated that 77% of the
30 listings agreed with the social security earnings report for the employer in any given year.
31 A comparison of city directory and death certificate information on occupations indicated a
December 1994 8-49 DRAFT-DO NOT QUOTE OR CITE
-------
1 match for occupation between at least one city directory listing and occupation on death
2 certificates for 68.3% of the study subjects.
3 This study demonstrated that city directories are a relatively inexpensive and accessible
4 source of occupational data for epidemiologic studies. Limitations of this study include the
5 problem in tracing women because of the change from maiden to married name and the
6 availability of data for only the year of residence in the city. They are superior to death
7 certificates in being able to identify high-risk employers in specific geographic sites.
8 Although death certificate data reflect usual lifetime occupation, city directories yield data on
9 short-term jobs, some of which may involve critical exposure. Thus, a combination of the
10 two approaches may be most productive in record-based hypothesis-generating studies. The
11 occupational data were missing for 15%, whereas employer data were missing for 36% in
12 the city directory. In the context of the mentioned pros and cons of using city directories,
13 this study found an excess risk for bladder cancer among two occupations with potential
14 diesel exposure: (1) truck drivers and (2) railroad workers. Two sources of bias that may
15 have influenced these findings are selection bias arising from the use of deaths instead of
16 incident cases, because survival for bladder cancer is high, and the absence of data on
17 confounding factors such as smoking, beverage consumption, and medication use.
18
19 8.4.5 Iscovich et al. (1987): Tobacco Smoking, Occupational Exposure,
20 and Bladder Cancer in Argentina
21 This is a hospital-based case-control study of bladder cancer conducted in La Plata,
22 Argentina, to estimate the risk of bladder cancer associated with different types of tobacco,
23 beverages, and occupational exposures. Bladder cancer is one of the most common cancers
24 among males in the La Plata area.
25 Cases were selected from patients with a histologically confirmed diagnosis of
26 bladder cancer (transitional, squamous-cell, or nonspecific cell type) admitted to the
27 10 general hospitals in the greater La Plata area (population in 1980 = 580,000) between
28 March 1983 and December 1985. Cases with true bladder papilloma and individuals who
29 were residents of greater La Plata for less than 5 years were excluded. Of the 120 cases
30 eligible to participate, 1 died prior to the interview, 2 refused to participate, and the
31 remaining 117 cases, representing 60% of the incident cases registered in the registry, were
December 1994 8-50 DRAFT-DO NOT QUOTE OR CITE
-------
1 interviewed. Two control groups (117 neighborhood and 117 hospital controls) were
2 matched by sex and age (±5 years). Of the 117 cases, 99 were males and 18 were females.
3 Hospital controls, selected from the same hospital as the cases, were hospitalized for the first
4 time within 3 mo of diagnosis of the illness of the cases. Twelve percent of the hospital
5 controls had illnesses known to be associated with tobacco smoking. Neighborhood controls
6 were sampled from among persons living in the same block. The interviewer proceeded
7 north in the block where the case resided and selected the first control who met the matching
8 criteria. Seven hospital controls could not be interviewed because of their poor physical
9 health and were replaced. Three neighborhood controls refused to participate and were
10 replaced. Cases and hospital controls were interviewed at the hospital and the neighborhood
11 controls at their homes to collect data on demographic, socioeconomic, and medical
12 variables, detailed smoking habits, and alcoholic and other beverages consumed.
13 The interviews were done by trained interviewers, two physicians, and a social worker.
14 The cases and hospital controls were interviewed in the hospital by the physicians; hence, the
15 interviews could not be conducted "blind". A detailed occupational history was obtained for
16 the three occupations of longest duration and the most recent one. For each job title, the
17 activity of the plant and type of production was also ascertained. Job titles were coded
18 according to the International Labor Union (ILO) 1970 classification. Plant activity and type
19 of production were coded according to the United Nations 1975 classification categories.
20 Information was also collected on exposure to 33 chemical and physical agents, which
21 included confirmed or suspected bladder carcinogens. A detailed history of smoking habits
22 was also obtained, and individuals were categorized as current smokers if they were smoking
23 at present or if they had stopped smoking less than 2 years previously. Exsmokers were
24 those who ceased smoking at least for 2 years or more than 2 years previously. For each
25 subject a cumulative lifelong consumption of cigarettes by type was estimated, and an
26 average consumption of cigarettes/day was computed.
27 Relative risks were computed for occupational factors using the unconditional logistic
28 regression method, adjusting for age and tobacco smoking. These risks were derived for
29 those who were ever employed in that occupation versus those who were never employed in
30 that occupation. Socioeconomic status of cases and neighborhood controls was similar but
31 there were fewer professionals and more manual workers among hospital controls compared
December 1994 8-51 DRAFT-DO NOT QUOTE OR CITE
-------
1 with cases. Occupational variables included job title and type of activity of the plant.
2 Significant excess risks were observed for truck and railroad drivers (RR = 4.31,
3 p < 0.002) and oil refinery workers (RR = 6.22, p < 0.02). The risk for truck and
4 railroad drivers was reduced after adjusting for smoking, whereas that for oil refinery
5 workers increased after adjusting for smoking (no RRs were presented). The adjusted
6 relative risks were not reported. A positive but nonsignificant association was observed for
7 printers (RR = 2.6, p < 0.77).
8 This study identified smoking and coffee drinking as the major risk factors for bladder
9 cancer in this area. The overall age-adjusted relative risk in males for current smokers
10 relative to nonsmokers was 7.2 (95% CI = 3.0, 20.1) with dose-response relationships
11 observed for the average daily amount as well as for duration of smoking. A strong dose-
12 response relationship was also observed for coffee drinking in males with a relative risk of
13 12 (95% CI = 4.3, 33.31) for those drinking more than three cups of coffee per day after
14 adjusting for the effect of smoking. No association was found with use of saccharin in
15 males. No results were presented for females for these risk factors.
16 This case-control study was conducted primarily to determine the reasons for the high
17 rates of bladder cancer in the La Plata region of Argentina. Only 60% of the cases
18 registered in the cancer registry were interviewed, and no information was provided for the
19 remaining 40% eligible nonrespondents to determine if the study sample was selectively
20 biased in any way. The sample size of 117 was small, and the analysis of males reduced it
21 to 99. Although the use of two different types of control groups is a strength of this study,
22 none of the interviews were done blind, and it appears that the hospital interviews were done
23 by the physicians and the neighborhood interviews were done by the social worker. Job
24 titles were used as surrogates of exposure, but the authors state that although they did attempt
25 to analyze by an exposure index derived from a job exposure matrix (details not provided),
26 they found no difference in exposure between cases and controls. This explanation is
27 ambiguous. The authors also grouped truck and railroad drivers together for reasons not
28 mentioned, and did not present separate risk estimates. A table showing the distribution of
29 cases and controls for selected activities or professions did not indicate if the data pertain to
30 both sexes or males only, and the text did not clarify that either. The reported significant
31 excess risks for truck and railroad drivers were reduced after adjusting for smoking, but it
December 1994 8-52 DRAFT-DO NOT QUOTE OR CITE
-------
1 was not known if the statistical significance persisted. No analysis was provided for the data
2 collected in the interviews on exposures to the 33 chemical and physical agents, and it was
3 not known if the truck and railroad drivers were operating diesel engines. Although rare in
4 the La Plata area, the authors acknowledge the occupations known to be associated with
5 bladder cancer (dye, rubber, leather, and textile workers).
6
7 8.4.6 Iyer et al. (1990): Diesel Exhaust Exposure and Bladder Cancer
8 Risk
9 This study is a hospital-based case-control study of bladder cancer and potential
10 exposure to diesel exhaust using data from a large ongoing case-control study of tobacco-
11 related neoplasms conducted by the American Health Foundation. An earlier study by
12 Wynder et al. (1985) looked at the relationship between occupational exposure to diesel
13 exhaust and the risk of bladder cancer. For this study, the objective was to evaluate the
14 relationship between the different measures of exposure to diesel exhaust, occupational and
15 self-reported, and the risk of bladder cancer. Cases comprised 136 patients with
16 histologically confirmed primary cancer of the urinary bladder interviewed at 18 hospitals in
17 six U.S. cities. Two controls were selected per case matched for sex, age, (within 2 years),
18 race, hospital, and year of interview. A total of 160 controls had nontobacco related
19 malignancies distributed as follows: stomach cancer (6%), colorectal cancer (20%), prostate
20 cancer (6%) and leukemia or lymphoma (8%). Among the 112 controls with nonmalignant
21 diseases, 3% had benign neoplasms, 6% had hyperplasis of the prostate, and 6% had
22 dorsopathies. Distribution of the other nonmalignant illnesses was not provided.
23 Occupational history included information on usual occupation and up to five other jobs.
24 Exposure to diesel exhaust in hobby activities also was collected. For the purpose of this
25 analysis, occupations were aggregated a priori into three categories: low probability of
26 exposure (reference group), possible exposure, and probable exposure. Analyses were also
27 done for self-reported exposure to diesel exhaust. Risk estimates were obtained by
28 unconditional logistic regression using PROC LOGIST of SAS. Cases and controls were
29 first compared by age, race, education, and smoking habit. Cases were found to be less
30 educated than controls (p < 0.05). Crude odds ratios for diesel exhaust exposure, based on
31 occupational or self-reported exposure, were not significantly elevated after controlling for
December 1994 8-53 DRAFT-DO NOT QUOTE OR CITE
-------
1 smoking and educational status (OR = 1.2, 95% CI = 0.8, 2.0). When individual
2 occupations were analyzed separately, truck drivers showed no excess risk (OR = 0.48,
3 95% CI = 0.15, 1.56).
4 The authors concluded that their study does not support the hypothesis of an association
5 between exposure to diesel exhaust and bladder cancer. Several significant limitations of
6 exposure assessment and analysis are evident in this study. In the introduction, the authors
7 stated that they refined the definition of exposure to diesel exhaust by obtaining a lifetime
8 occupational history, but in the methods section they stated that they restricted analysis to
9 usual occupational history and five other jobs, which was not that different from their earlier
10 study (Wynder et al., 1985). The terms, low probability of exposure, possible exposure, and
11 probable exposure also were not clearly defined. Information on duration of employment or
12 exposure was not presented, and no attempts were made to validate occupational history.
13 No information was available on calendar years of employment in the truck-driving industry
14 or the locomotive occupations. Because diesel trucks and locomotives were introduced in the
15 mid 1950s and the dieselization was completed by 1960, it would be important to use 1960
16 as a cut-off date and to restrict analysis to subjects who worked in these industries after that
17 ,date. No information on nonrespondent cases and controls was provided. The authors
18 indicated hi the method section that cases were individually matched to controls, but they
19 performed an unmatched analysis to calculate the odds ratios and do not address why they
20 did not preserve the matching in the analysis, especially because such an analysis could bias
21 the risk estimates to unity.
22
23 8.4.7 Steineck et al. (1990): Increased Risk of Urothelial Cancer in
24 Stockholm from 1985 to 1987, After Exposure to Benzene and
25 Exhausts
26 This study was conducted to investigate the association between benzene, diesel, and
27 petrol exhausts as well as some other industry-related chemicals and the risk of urothelial
28 cancer. Designed as a population-based case-control study, it was conducted among all men
29 born between 1911 and 1945 and living in the County of Stockholm for all or part of the
30 observation period (September 15, 1985, to November 30, 1987). All incident cases of
31 urothelial cancer and squamous-cell carcinoma of the lower urinary tract were contacted for
32 inclusion in the study. Controls were selected by stratified random sampling during the
December 1994 8-54 DRAFT-DO NOT QUOTE OR CITE
-------
1 observation period from a computerized register for the population of Stockholm. A postal
2 questionnaire was sent to study subjects at their homes to collect information on occupational
3 history. The questions on occupation included exposure to certain specified
4 occupations/industries/chemicals and lists of all jobs held and duration in each job.
5 An industrial hygienist, unaware of case-control status, classified subjects as being exposed
6 or unexposed to 38 agents and groups of substances including 17 exposure categories with
7 aromatic amines. Using all the exposure information, the exposure period was defined and
8 the annual dose was rated as low, moderate, or high based on the accumulated dose
9 (exposure duration multiplied by intensity of exposure) during the course of 1 average year
10 for the defined exposure period. Swedish and international data were used to classify
11 subjects as exposed, based on air concentrations in the work environment that were higher
12 than for the general public, or skin contact with liquids of low volatility. To allow for
13 latency, the authors ignored exposures after 1981. Data were gathered from 256 cases and
14 287 controls. Controls were selected by stratified random sampling four times from the
15 computerized register during the observation period of the population of the County of
16 Stockholm. These subjects comprised 80% of eligible cases and 79% of eligible controls.
17 Nine of the cases and 16% of the controls refused to participate in the study. The
18 distribution of urothelial cancers was as follows: 5 tumors in the renal pelvis, 243 in the
19 urinary bladder, 5 in the ureter, none in the urethra, and 3 at multiple sites. Two cases who
20 were exposed to a high annual dose of aromatic amines were omitted from all further
21 analysis to eliminate their confounding effects. Crude relative risks were calculated for men
22 classified as exposed or not exposed to several substances. Twenty-five cases and
23 19 controls reported having been exposed to diesel exhaust, yielding an odds ratio of
24 1.7 (95% CI = 0.9, 3.3). The corresponding relative odds for petrol exhausts based on
25 24 cases and 24 controls were 1.0 (95% CI = 0.5, 1.9). Odds ratios were then calculated
26 for low, moderate, and high levels of the annual dose adjusted for smoking and year of birth.
27 For diesel exhausts, the odds ratio for low levels was 1.3 (95% CI = 0.6, 3.1), for moderate
28 levels was 2.2 (95% CI = 0.7, 6.6), and for high levels was 2.9 (95% CI = 0.3, 30.0)
29 indicating a dose response. The corresponding odds ratios for petrol exhausts were
30 0.6 (95% CI = 0.3, 1.3), 1.4 (95% CI = 0.5, 3.7), and 3.9 (95% CI = 0.4, 35.5).
31 Restricting the analysis to only moderate or high annual doses of exposure adjusted for year
December 1994 8-55 DRAFT-DO NOT QUOTE OR CITE
-------
1 of birth and smoking showed a sevenfold increased risk for subjects exposed to both diesel
2 and petrol exhausts (OR = 7.1, 95% CI = 0.9, 58.8). For exposure to diesel (OR =1.1)
3 and petrol (OR = 1.0) exhausts alone, no excess risk was detected in this analysis. Odds
4 ratios were calculated for low, moderate, and high exposure to benzene with rates of
5 1.7 (95% CI = 0.6, 5.1) for low annual doses, 1.1 (95% CI = 0.3, 4.5) for moderate
6 annual doses, and 3.0 (95% CI = 1.0, 8.7) for high annual doses.
7 The authors discuss misclassification and confounding as sources of bias in this study.
8 To examine misclassification they compared hygienist-assessed exposure and self-reported
9 exposure for printing ink and found a higher relative risk and fewer exposed subjects for
10 hygienist-assessed exposure indicating that specificity was a problem for self-reported
11 exposure. It is not known to what extent this may have affected the risk estimates for diesel
12 exhausts since data on self-reported exposure to diesel are not presented. They also mention
13 the possibility of exposure misclassification from using an average annual dose in which a
14 person exposed to an agent at a high level for a few working days and a person exposed to a
15 low level for many days are both rated as exposed to low annual doses. Although
16 statistically nonsignificant elevated odds ratios of 1.3, 2.3, and 2.9 were derived for low,
17 moderate, and high levels of diesel exposure, the authors state that some of their subjects
18 may have later worked in jobs with benzene exposure, and because an elevated risk was
19 detected for benzene exposure, this confounding effect may explain some of the excess risk.
20 An almost statistically significant interaction was observed for exposure to combined diesel
21 and petrol exhausts (OR = 7.1, 95% CI = 0.9, 58.8), which changed to
22 5.1 (95% CI = 0.6, 43.3) after adjustment for benzene exposure, again providing evidence
23 for the confounding role of benzene exposure in explaining some of the observed results.
24 Table 8-3 summarizes the bladder cancer case-control studies.
25
26
27 8.5 DISCUSSION AND SUMMARY
28 Certain extracts of diesel exhaust have been demonstrated to be both mutagenic and
29 carcinogenic in animals and in humans. Animal data are suggestive that diesel exhaust is a
30 pulmonary carcinogen among rodents exposed by inhalation to high doses over long periods
December 1994 8-56 DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 8-3. EPIDEMIOLOGIC STUDIES OF THE HEALTH EFFECTS OF EXPOSURE TO DIESEL EXHAUST:
CASE-CONTROL STUDIES OF BLADDER CANCER
Authors
Population Studied
Diesel Exhaust Exposure
Assessment
Results
Limitations
Howe et al. 480 Male case-control pairs
(1980)
152 Female case-control pairs
Cases diagnosed between April
1974 and June 1976 in three
Canadian provinces
Matched on age and sex
Based on occupational SNS RR = 2.8 for diesel Exposure based on occupational history, which
history of jobs involving and traffic fumes
exposure to dust and
fumes
A priori suspect
industries
SS RR = 9.00 for
railroad workers
was not validated
Diesel exhaust and traffic fumes were combined.
Only 77% of eligible population included in the
study
oo
Wynder et al. 194 Histologically confirmed
(1985) male cases between the ages of
20-80 years
582 Matched controls (age,
race, year of interview, and
hospital of admission); diseases
not related to tobacco use
From 18 hospitals located in
six U.S. cities between
January 1981 and May 1983
Occupational titles were
defined by Industrial
Hygiene Standard into
dichotomous "exposed"
and "not exposed".
Also defined by NIOSH
standards into "high
exposure", "moderate
exposure", and "low
exposure"
SNS ORs were 1.68 and
0.16 for high and
moderate exposure,
respectively, as compared
to low exposure.
Exposure based on usual occupation may have
lead to misclassification.
Dichotomous classification made dose-response
analysis unattainable.
No data on other confounders such as smoking
-------
I
n>
TABLE 8-3 (cont'd). EPIDEMIOLOGIC STUDIES OF THE HEALTH EFFECTS OF EXPOSURE TO DIESEL
EXHAUST: CASE-CONTROL STUDIES OF BLADDER CANCER
Authors
Population Studied
Diesel Exhaust Exposure
Assessment
Results
Limitations
oo
10 years of employment
Positive trend
(p = 0.006) observed
with increasing duration
of employment as truck
driver
Exposure defined as occupation of "truck driver"
(i.e., it could have been diesel or gasoline or
both)
No histogical confirmation of bladder cancer
diagnosis
No data on other confounders such as other
exposures, smoking, etc.
Steenland
et al. (1987)
648 Male bladder cancer deaths
from Hamilton County, OH
1,275 Matched controls from
Occupation or industry
listed in city directory
and on death
certificates.
OR = 12 (p = 0.01)
for truck drivers with
>20 years of
employment
Exposure based on city directory or death
certificate listing that was not validated.
Lack of controlling for confounders
other deaths (pool of six controls
for each case, excluding urinary
tract tumors and pneumonias
matched on sex, age at death,
year of death, race)
OR = 2.21 (p < 0.05)
for railroad workers
with >20 years of
employment
City directory usually has short-term job listing
Missing data on 15% of occupations and 36%
for employers hi the directory
n
H
W
-------
I
VO
VO
TABLE 8-3 (cont'd). EPIDEMIOLOGIC STUDIES OF THE HEALTH EFFECTS OF EXPOSURE TO DIESEL
EXHAUST: CASE-CONTROL STUDIES OF BLADDER CANCER
Authors
Population Studied
Diesel Exhaust Exposure
Assessment
Results
Limitations
oo
ISl
Iscovich 117 Histologically confirmed
et al. (1987) bladder cancer cases (60% of
all incident cases)
117 Hospital controls and
117 neighborhood controls
(matched on age and sex)
Cses and hospital controls from
10 general hospitals in greater
La Plata between March 1983
and December 1985
Past and present
occupational data
were collected by
questionnaire.
An exposure index based
on a job exposure matrix
was generated.
SS OR = 4.3 for truck
and railway drivers
SS RR = 6.2 for oil
refinery workers
Exposure based on job held that was not
validated
40% of eligible cases were nonrespondent.
Small sample size
Interviewers were not "blind" to the status of an
individual, and this could have biased the
findings.
Truck and railroad drivers were grouped
together.
Not adjusted for other confounding exposures
such as dye, rubber, etc.
o
o
o
H
O
1
m
o
n
a
Iyer et al. 136 histologically confirmed
(1990) bladder cancer cases
Lifetime occupational
history
No excess found.
272 controls, two each matched Self-reported diesel
on sex, age, race, hospital, and exhaust exposure
year of interview
(160 malignant, 112
nonmalignant)
From 18 hospitals in six U.S.
cities
Exposure aggregated a
priori into:
Low probability
Possible
Probable
Exposure based on self report, which was not
validated
Although lifetime occupational history was
obtained, analysis was restricted to usual
occupation.
A priori classification was ambiguous.
-------
TABLE 8-3 (cont'd). EPIDEMIOLOGIC STUDIES OF THE HEALTH EFFECTS OF EXPOSURE TO DIESEL
EXHAUST: CASE-CONTROL STUDIES OF BLADDER CANCER
Population Studied
Authors
Steineck Population based study from
et al. (1990) County of Stockholm
Men born between 1911 and
1945
256 (243 bladder) urinary tract
cancer incident cases (80% of
eligibles)
287 Controls (79% of eligibles)
from population of Stockholm
Observation period September 15,
1985, to November 30, 1987
Diesel Exhaust Exposure
Assessment
Results
Limitations
oo
o\
o
Occupational history
classified into exposed
and nonexposed by
industrial hygienist
"blind" towards case or
control status
Using all exposure
information, annual
dose rated as "low",
"moderate", and "high"
SNSOR = 1.3 for low,
OR = 2.2 for moderate,
and OR = 2.9 for high
exposure were observed
for diesel exposure.
SNSOR = 7.1 observed
for diesel and gasoline
exhaust combined
exposure
Elaborate exposure history classification not used
to advantage by simultaneous adjustment
Misclassification in exposure may have occurred
Small sample size of only 25 cases and
19 controls were exposed to diesel exhaust.
Confounding by other exposures not accounted
for, except benzene
Abbreviations:
OR = Odds ratio.
RR = Relative risks.
SNS = Statistically nonsignificant.
SS = Statistically significant.
s
o
I
n
H
w
-------
1 of time. Because large working populations are currently exposed to diesel exhaust and
2 because nonoccupational ambient exposures currently are of concern as well, the possibility
3 that exposure to this complex mixture may be carcinogenic to humans has become an
4 important public health issue.
5 Because diesel emissions become diluted in the ambient air, it is difficult to study the
6 health effects in the general population. Nonoccupational exposure to diesel exhaust is
7 worldwide in urban areas. Thus, "unexposed" reference populations used in occupational
8 cohort studies are likely to contain a substantial number of individuals who are
9 nonoccupationally exposed to diesel exhaust. Furthermore, the "exposed" group in these
10 studies is based on job titles which in most instances are not verified or correlated with
11 environmental hygiene measurement. The issue of health effect measurement gets further
12 complicated by the fact that occupational cohorts tend to be healthy and have below-average
13 mortality, usually referred as the "healthy worker effect". Hence, the usual standard
14 mortality ratios observed in cohort mortality studies are underestimations of real risk.
15 A major difficulty with the occupational studies considered here was the measurement
16 of the actual diesel exhaust exposure. Because all the cohort mortality studies were
17 retrospective, the assessment of health effects from exposure to diesel exhaust was naturally
18 indirect. In these occupational settings, no systematic quantitative records of ambient air
19 were available. Most studies compared men in job categories with presumably some
20 exposure to diesel exhaust with either standard populations (presumably no exposure to diesel
21 exhaust) or with men in other job categories from industries with little or no potential for
22 diesel exhaust exposure. A few studies have included measurements of diesel fumes, but
23 there is no standard method for the measurement. No attempt is made to correlate these
24 exposures with the cancers observed in any of these studies, nor is it clear exactly which
25 extract should have been measured to assess the occupational exposure to diesel exhaust. All
26 studies have relied on the job categories or self report of exposure to diesel exhaust. In the
27 studies by Garshick et al. (1987, 1988), the diesel exhaust-exposed job categories were
28 verified based on an industrial hygiene survey done by Woskie et al. (1988a,b). It was
29 found by the investigators that in most cases the job titles were good surrogates for diesel
30 exhaust exposure. Also, in this railroad industry where only persons who had at least
31 10 years of work experience were included in the study, the workers tended not to change
December 1994 8-61 DRAFT-DO NOT QUOTE OR CITE
-------
1 job categories over the years. Thus, a job known only at one point in tune was a reasonable
2 marker of past diesel exhaust exposure. Unfortunately the exposure was only qualitatively
3 verified. The quantitative use of this information would have been much more meaningful.
4 The occupations involving potential exposure to diesel exhaust are miners, truck drivers,
5 transportation workers, railroad workers, and heavy equipment operators.
6 With the exception of the study by Waxweiler et al. (1973), there have been no known
7 studies of miners to assess whether diesel exhaust is associated with lung cancer. Currently,
8 there are about 385 underground metal mines in the United States. Of these, 250 have been
9 permanently operating, whereas 135 have been intermittently operating (Steenland, 1986).
10 Approximately 20,000 miners are employed, but all of them are not currently working in the
11 mines. Diesel engines were introduced in the metal mines in the early to mid 1960s.
12 Although all these mines use diesel equipment, it is difficult to estimate how many of these
13 miners were actually exposed to diesel fumes.
14 Diesel engines were introduced in coal mines at an even later date, and their use is still
15 quite limited. In 1983, approximately 1,000 diesel units were in place in underground coal
16 mines, up from about 200 units in 1977 (Daniel, 1984). The number of units per mine
17 varies greatly; one mine may account for over 100 units.
18 Even if it were possible to estimate how many miners (metal and coal) were exposed to
19 diesel exhaust, it would be very difficult to separate out the confounding effects of other
20 potential pulmonary carcinogens, such as radon decay products, heavy metals (such as
21 arsenic, chromium), etc. Furthermore, the relatively short latency period limits the
22 usefulness of these cohorts of miners.
23
24 8.5.1 The Cohort Mortality Studies
25 The cohort studies mainly demonstrated an increase in lung cancer. Studies of bus
26 company workers by Waller (1981), Rushton et al. (1983), and Edling et al. (1987) failed to
27 demonstrate any statistically significant excess risk of lung cancer, but these studies have
28 certain methodological problems, such as small sample sizes, short follow-up periods (just
29 6 years in the Rushton et al. study), lack of information on confounding variables, and lack
30 of analysis by duration of exposure, duration of employment, or latency, that preclude their
31 use in determining the carcinogenicity of diesel exhaust. Although the Waller (1981) study
December 1994 8-62 DRAFT-DO NOT QUOTE OR CITE
-------
1 had a 25-year follow-up period, the cohort was restricted to only employees (ages 45 to 64)
2 currently in service. Employees who left the job earlier, as well as those who were still
3 employed after age 64 and who may have died from cancer, were excluded.
4 Wong et al. (1985) conducted a mortality study of heavy equipment operators that
5 demonstrated a significant increased risk of liver cancer in total and in various subcohorts.
6 The same analysis also showed statistically significant deficits in cancers of the large intestine
7 and rectum. Metastasis from the cancers of the large intestine and rectum in the liver
8 probably were misclassified as primary liver cancer which led to an observed excess risk.
9 This study did demonstrate a nonsignificant positive trend for cancer of the lung with length
10 of membership and latency. Analysis of deceased retirees showed a significant excess of
11 lung cancer. Individuals without work histories who started work prior to 1967 when
12 records were not kept may have been in the same jobs for the longest period of time.
13 Workers without job histories included those who had the same job before and after 1967 and
14 thus may have worked about 12 to 14 years longer; these workers exhibited significant
15 excess risks of lung cancer and stomach cancer. If this assumption about duration of jobs is
16 correct, then these site-specific causes can be linked to diesel exhaust exposure. One of the
17 methodological limitations of this study is that most of these men worked outdoors; thus, this
18 cohort might have had relatively low exposure to diesel exhaust. The authors did not present
19 any environmental measurement data either. Because of the absence of detailed work
20 histories for 30% of the cohort and the availability of only partial work histories for the
21 remaining 70%, jobs were classified and ranked according to presumed diesel exposure.
22 Information is lacking regarding duration of employment in the job categories (used for
23 surrogate of exposure) and other confounding factors (alcohol consumption, cigarette
24 smoking, etc.). Thus, this study cannot be used to support a causal association or to refute
25 the same between exposure to diesel exhaust and lung cancer.
26 A 2-year mortality analysis by Boffetta and Stellman (1988), of the American Cancer
27 Society's prospective study, after controlling for age and smoking, demonstrated an excess
28 risk of lung cancer in certain occupations with potential exposure to diesel exhaust. These
29 excesses were statistically significant among miners (RR = 2.67, 95% CI = 1.63, 4.37) and
30 heavy equipment operators (RR = 2.6, 95% CI = 1.12, 6.06). The elevated risks were
31 nonsignificant in railroad workers (RR = 1.59) and truck drivers (RR = 1.24). A dose
December 1994 8-63 DRAFT-DO NOT QUOTE OR CITE
-------
1 response was also observed for truck drivers. With the exception of miners, exposure to
2 diesel exhaust occurred in the three other occupations showing an increase in the risk of lung
3 cancer. Despite methodologic limitations such as the lack of representiveness of the study
4 population (composed of volunteers only who were probably healthier than the general
5 population) leading to an underestimation of the risk and the questionable reliability of
6 exposure data based on self-administered questionnaires that were not validated, this study is
7 suggestive of a causal association between exposure to diesel exhaust and excess risk of lung
8 cancer.
9 Two mortality studies of railroad workers were conducted by Howe et al. (1983) in
10 Canada and Garshick et al. (1988) in the United States. The Canadian study found relative
11 risks of 1.2 (p < 0.01) and 1.35 (p < 0.001) among "possibly" and "probably" exposed
12 groups, respectively. The trend test showed a highly significant dose-response relationship
13 with exposure to diesel exhaust and the risk of lung cancer. The main limitation of the study
14 was the inability to separate overlapping exposures of coal dust and diesel fumes.
15 Information on jobs was available at retirement only. There was also insufficient detail on
16 the classification of jobs by diesel exhaust exposure. The exposures could have been
17 nonconcurrent or concurrent, but since the data are lacking, it is possible that the observed
18 excess could be due to the effect of both coal dust and diesel fumes and not due to just one
19 or the other. However, it should be noted that, so far, coal dust has not been demonstrated
20 to be a pulmonary carcinogen in studies of coal miners, but lack of data on confounders such
21 as asbestos and smoking makes interpretation of this study difficult. When three diesel
22 exhaust exposure categories were examined for smoking-related diseases such as emphysema,
23 laryngeal cancer, esophageal cancer, and buccal cancer, positive trends were observed,
24 raising a possibility that the dose-response demonstrated for diesel exposure may have been
25 due to smoking. The findings of this study are at best suggestive of diesel exhaust being a
26 lung carcinogen.
27 The most definitive evidence for linking diesel exhaust exposure to lung cancer comes
28 from the Garshick et al. (1988) railroad worker study conducted in the United States.
29 Relative risks of 1.57 (95% CI = 1.19, 2.06) and 1.34 (95% CI = 1.02, 1.76) were found
30 for ages 40 to 44 and 45 to 49, respectively, after the exclusion of workers exposed to
31 asbestos. This study also found that the risk of lung cancer increased with increasing
December 1994 8-64 DRAFT-DO NOT QUOTE OR CITE
-------
1 duration of employment. As this was a large cohort study with lengthy follow-up and
2 adequate analysis, including dose response (based on duration of employment as a surrogate)
3 as well as adjustment for other confounding factors such as asbestos, the observed association
4 between increased lung cancer and exposure to diesel exhaust is more meaningful.
5
6 8.5.2 Case-Control Studies of Lung Cancer
7 Among the eight lung cancer case-control studies reviewed in this chapter, only one,
8 the study by Lerchen et al. (1987), did not find an increased risk of lung cancer, after
9 adjusting for age and smoking, for diesel fume exposure. The major limitation of this study
10 was a lack of adequate exposure data derived from the job titles obtained from occupational
11 histories. Next of kin provided the occupational histories for 50% of the cases that were not
12 validated. The power of the study was small (analysis done on males only, 333 cases).
13 On the other hand, statistically nonsignificant excess risks were observed for diesel exhaust
14 exposure by Williams et al. (1977) in railroad workers (OR =1.4) and truck drivers
15 (OR = 1.34), by Hall and Wynder (1984) in workers who were exposed to diesel exhaust
16 versus those who were not (OR =1.4 and 1.7 with two different criteria), and by Damber
17 and Larsson (1987) in professional drivers (OR = 1.2). These rates were adjusted for age
18 and smoking. Both Williams et al. (1977) and Hall and Wynder (1984) had high
19 nonparticipation rates of 47 and 36%, respectively. Therefore, the positive results found in
20 these studies are underestimated at the best. In addition, the self-reported exposures used in
21 the study by Hall and Wynder (1984) were not validated. This study also had low power to
22 detect excess risk of lung cancer for specific occupations.
23 The study by Benhamou et al. (1988), after adjusting for smoking, found significantly
24 increased risks of lung cancer among French motor vehicle drivers (RR = 1.42) and
25 transport equipment operators (RR = 1.35). The main limitation of the study was the
26 inability to separate the exposures to diesel exhaust from those of gasoline exhausts since
27 both motor vehicle drivers and transport equipment operators probably were exposed to the
28 exhausts of both types of vehicles.
29 Hayes et al. (1989) combined data from three studies (conducted in three different
30 states) to increase the power to detect an association between lung cancer and occupations
31 with a high potential for exposure to diesel exhaust. They found that truck drivers employed
December 1994 8-65 DRAFT-DO NOT QUOTE OR CITE
-------
1 for more than 10 years had a significantly increased risk of lung cancer (OR =1.5,
2 95% CI = 1.1, 1.9). This study also found a significant trend of increasing risk of lung
3 cancer with increasing duration of employment among truck drivers. The relative odds were
4 computed by adjusting for birth cohort, smoking, and state of residence. The main limitation
5 of this study is again the mixed exposures to diesel and gasoline exhausts, since information
6 on type of engine was lacking. Also, potential bias may have been introduced because the
7 way in which the cause of death was ascertained for the selection of cases varied in the three
8 studies. Further, the methods used in these studies to classify occupational categories were
9 different probably leading to incompatibility of occupational categories.
10 The most convincing evidence comes from the Garshick et al. (1987) case-control study
11 of railroad workers and the Steenland et al. (1990) case-control study of truck drivers in the
12 Teamsters Union. Garshick et al. found that after adjustment for asbestos and smoking, the
13 relative odds for continuous exposure were 1.39 (95% CI = 1.05, 1.83). Among the
14 younger workers with longer diesel exhaust exposure, the risk of lung cancer increased with
15 the duration of exposure after adjusting for asbestos and smoking. Even after the exclusion
16 of recent diesel exhaust exposure (5 years before death), the relative odds increased to
17 1.43 (95% CI = 1.06, 1.94). This study appears to be a well-conducted and well-analyzed
18 case-control study with reasonably good power. Potential confounders were controlled
19 adequately, and interactions between diesel exhaust and other lung cancer risk factors were
20 tested.
21 Steenland et al. (1990), on the other hand, created two separate work history files, one
22 from Teamsters Union pension files and the other from next-of-kin interviews. Using
23 duration of employment as a categorical variable and considering employment after 1959
24 (when presumed dieselization occurred) for long haul drivers, the risk of lung cancer
25 increased with increasing years of exposure. Using 1964 as the cut-off, a similar trend was
26 observed for long haul drivers. For short haul drivers the trend was positive with a 1959
27 cut-off but not when 1964 was used as the cut-off. For truck drivers who primarily drove
28 diesel trucks and worked for 35 years, the relative odds were 1.89. The limitations of this
29 study include possible misclassifications of exposure and smoking, lack of levels of diesel
30 exposure, smaller nonexposed group, and insufficient latency period. Given these
31 limitations, the findings of this study are probably underestimated.
December 1994 8-66 DRAFT-DO NOT QUOTE OR CITE
-------
1 8.5.3 Case-Control Studies of Bladder Cancer
2 Of the seven bladder cancer case-control studies, four studies found increased risk in
3 occupations with a high potential diesel exhaust exposure. A significantly increased risk of
4 bladder cancer was found in Canadian railroad workers (RR = 9.0, 95% CI = 1.2, 349.5;
5 Howe et al., 1980), truck drivers from New Hampshire and Vermont (OR = 2.9, p < 0.05;
6 Hoar and Hoover, 1985), and in Argentinean truck and railroad drivers (RR = 4.31,
7 p < 0.002; Iscovich et al., 1987). A positive trend with increasing employment as truck
8 driver (p = 0.006) was observed by Hoar and Hoover, 1985 in their study of truck drivers
9 from New Hampshire and Vermont. Significantly increased risks also were observed with
10 increasing duration of employment of >20 years in truck drivers (OR = 12, p = 0.01) and
11 railroad workers (OR = 2.21, p< 0.05; Steenland et al., 1987). No significant increased
12 risk was found for any diesel-related occupations in studies by Wynder et al. (1985), Iyer
13 et al. (1990), or Steineck et al. (1990). All these studies had several limitations including
14 inadequate characterization of diesel exhaust exposure, lack of validation of surrogate
15 measures of exposure, and presence of other confounding factors (cigarette smoking, urinary
16 retention, concentrated smoke within the truck cab, etc.); most of them had small sample
17 sizes, and none presented any latency analysis.
18
19 8.5.4 Relevant Methodologic Issues
20 Throughout this chapter various methodologic limitations of individual studies have
21 been discussed, such as small sample size, short follow-up period, lack of latency analysis,
22 and lack of data on confounding variables. However, two of the major methodologic
23 concerns in these studies are use of death certificates to determine the cause of death and
24 lack of data on cigarette smoking which is a strong risk factor for both lung cancer and
25 bladder cancer. Death certificates were used by all of the seven cohort mortality studies, two
26 case-control studies of lung cancer, and one case-control study of bladder cancer, to
27 determine cause of death. Use of death certificates could lead to misclassification bias.
28 Studies of autopsies done between 1960 and 1971 demonstrated that lung cancer was
29 overdiagnosed when compared to hospital discharge with no incidental cases found at autopsy
30 (Rosenblatt et al., 1971). Schottenfeld et al. (1982) also found an overdiagnosis of lung
31 cancer among autopsies conducted in 1977 and 1978. On the other hand, Percy et al. (1981)
December 1994 3.57 DRAFT-DO NOT QUOTE OR CITE
-------
1 noted 95% concordance when comparing 10,000 lung cancer deaths observed in the Third
2 National Cancer Survey during 1969 to 1971 (over 90% were confirmed histologically) to
3 death certificate-coded cause of death. For bladder cancer, the concordance rate was 91%.
4 These more recent findings suggest that the diagnosis of lung cancer as well as bladder
5 cancer on death certificates is better than anticipated. Furthermore, an overdiagnosis of lung
6 cancer or bladder cancer on death certificates would reduce the ability of the study to detect
7 an effect of diesel exhaust exposure.
8 All the cohort studies considered for this report are retrospective mortality studies. It is
9 usually difficult to obtain smoking history in such instances. The smoking histories obtained
10 from surrogates (next of kin being either a spouse or an offspring) were found to be accurate
11 by Lerchen and Samet (1986) and McLaughlin et al. (1987). Lerchen and Samet (1986) did
12 not detect any consistent bias in the report of cigarette consumption. In contrast,
13 overreporting of cigarette smoking by surrogates was observed by Rogot and Reid (1975),
14 Kolonel et al. (1977), and Humble et al. (1984). Kolonel et al. (1977) found that the age at
15 which an individual started smoking was reported within 4 years of actual age 84% of the
16 time. The indication from these studies is that surrogates were able to provide fairly credible
17 information on the smoking habits of the study subjects. If the surrogates of the cases were
18 more likely to overreport cigarette smoking as compared to the controls, then it might be
19 harder to find an effect of diesel exhaust because most of the increase in lung cancer would
20 be attributed to smoking rather than to the effect of exposure to diesel exhaust.
21
22 8.5.5 Criteria of Causal Inference
23 In most situations epidemiologic data are used to delineate the causality of certain
24 health effects. Several cancers have been causally associated with exposure to agents for
25 which there is no direct biological evidence. Insufficient knowledge about the biological
26 basis for diseases in humans makes it difficult to identify exposure to an agent as causal,
27 particularly for malignant diseases when the exposure was in the distant past. Consequently,
28 epidemiologists and biologists have provided a set of criteria that define a causal relationship
29 between exposure and the health outcome. A causal interpretation is enhanced for studies
30 that meet these criteria. None of these criteria actually proves causality; actual proof is
31 rarely attainable when dealing with environmental carcinogens. None of these criteria should
December 1994 g-68 DRAFT-DO NOT QUOTE OR CITE
-------
1 be considered either necessary (except temporality of exposure) or sufficient in itself. The
2 absence of any one or even several of these criteria does not prevent a causal interpretation.
3 However, if more criteria apply it provides credible evidence for causality.
4 Thus, applying the criteria of causal inference to the seven cohort mortality and eight
5 case-control studies in which risk of lung cancer was assessed, resulted in the following:
6 • Temporality: There is a temporality of exposure to diesel exhaust prior to the
7 occurrence of lung cancer.
8
9 • Strength of Association: The strength of association between exposure and the
10 occurrence of lung cancer in the cohort studies showed a 30 to 57% higher risk
11 among exposed as compared to nonexposed (Howe et al., 1983; Wong et al.,
12 1985; Boffetta and Stellman, 1988; Garshick et al., 1988). In case-control studies,
13 the risk varied from 20 to 89% higher among exposed as compared to nonexposed
14 (Williams et al., 1977; Hall and Wynder, 1984; Damber and Larsson, 1987;
15 Garshick et al., 1987; Benhamou et al., 1988; Hayes et al., 1989; Steenland et al.,
16 1990). Some of these studies did adjust for the confounding effects of smoking,
17 asbestos, and other exposures.
18
19 • Consistency: Four cohort studies and seven case-control studies of lung cancer
20 conducted in several populations in the United States and Europe consistently.
21 found the same effect (i.e., lung cancer).
22
23 • Specificity: All of the above-mentioned studies found the same specific effect
24 (i.e., lung cancer).
25
26 • Biological Gradient: The biological gradient, which refers to the dose-response
27 relationship, was observed in the cohorts of Canadian railway workers (Howe
28 et al., 1983), heavy bulldozer operators (Wong et al., 1985), and truck drivers
29 who had enrolled in the American Cancer Society's prospective mortality study
30 (Boffetta and Stellman, 1988). In the case-control studies, a dose response was
31 observed in railroad workers (Garshick et al., 1988; Hayes et al., 1989; Steenland
32 et al., 1990). Although other studies failed to observe a dose response, these
33 studies were methodologically limited due to confounding by other exposures and
34 lack of either quantitative data on exposure or surrogate data on dose.
35
36 • Biological Plausibility: Because diesel exhaust particles consist of a carbon core
37 with surface layers of organics, the tumorigenic activity either resides in one or
38 both of these components. As explained in Chapter 9, there is clear evidence that
39 the organic constituents have the capacity to interact with DNA and give rise to
40 mutations, chromosomal aberrations, and cell transformations, all well-established
41 steps in the process of carcinogenesis. Furthermore, these organic chemicals
42 include a variety of polycyclic aromatic hydrocarbons (PAHs) and nitroaromatics,
43 many of which are known to be pulmonary carcinogens. Alternatively, Vostal
44 (1986) suggests that "diesel" particles themselves induce lung cancer, most likely
December 1994 8-69 DRAFT-DO NOT QUOTE OR CITE
-------
1 via an epigenetic mechanism, if they are present at sufficiently high doses. This
2 makes a convincing argument for biological plausibility of lung cancer occurrence.
3
4 When the same causal inference criteria were applied to the seven case-control
5 studies in which risk of bladder cancer was assessed, the results were:
6 • Temporality: There is temporality of exposure to diesel exhaust prior to the
7 occurrence of bladder cancer.
8
9 • Strength of Association: The relative odds of getting bladder cancer among
10 exposed as compared to nonexposed ranged from 2 to 12 times higher (Howe
11 et al., 1980; Hoar and Hoover, 1985; Iscovich et al., 1987; Steenland et al.,
12 1987). None of these studies adjusted for other confounding effects such as
13 cigarette smoking, exposures to other chemicals, urinary retention, etc.
14
15 • Consistency: Four out of seven bladder case-control studies conducted in the
16 United States and abroad found an increased relative odds of bladder cancer in the
17 exposed population. None of the cohort studies showed increased bladder cancer
18 mortality; however, people rarely die from bladder cancer, so bladder cancer
19 excess is unlikely to be detected in mortality studies.
20
21 • Specificity: Four out of seven case-control studies found an excess of bladder
22 cancer. The specificity criterion, per se, does not apply in this particular instance
23 since these are case-control studies.
24
25 • Biological Gradient: Dose response was observed in two out of four studies
26 showing increasing relative odds with increasing length of employment (Hoar and
27 Hoover, 1985; Steenland et al., 1987).
28
29 • Biological Plausibility: It has been demonstrated that motor exhaust emissions
30 contain PAHs and nitro-PAHs (Stenberg et al., 1983; Rosenkranz and
31 Mermelstein, 1983). There is some evidence that nitro-PAHs may be responsible
32 for the induction of human bladder cancer. Nitro-PAHs can be metabolized to
33 aromatic amine derivatives, and some of these agents are known to be capable of
34 inducing urinary bladder cancer (Clayson and Garner, 1976). Furthermore,
35 1-nitropyrene (1-NP) has been reported to be carcinogenic in the rat mammary
36 gland (Hirose et al., 1984); the structurally related 4-aminobiphenyl, which induces
37 bladder cancer in humans, also induces mammary gland tumors in rats (Hirose
38 et al., 1984). Although the applicability of these experimental results to humans is
39 unknown, the laboratory evidence certainly suggests the biological plausibility of
40 diesel exhaust to be a urinary bladder carcinogen.
41
42 In summary, although some of the causality inference criteria do apply to bladder
43 cancer, the evidence for bladder cancer in populations exposed to diesel exhaust is
44 inadequate. On the other hand, all the causality inference criteria apply well for lung cancer.
December 1994 g-70 DRAFT-DO NOT QUOTE OR CITE
-------
1 An excess risk of lung cancer was observed in four out of seven cohort studies and seven out
2 of eight case-control studies. Dose response was found in three cohort studies and three
3 case-control studies using duration of employment as a surrogate for dose. However,
4 because of lack of the actual data on exposure to diesel exhaust in these studies and other
5 methodologic limitations, such as insufficient latency for lung cancer to develop, the human
6 evidence falls short of being sufficient, and hence is considered to be limited for diesel
7 exhaust exposure.
December 1994 g_71 DRAFT-DO NOT QUOTE OR CITE
-------
1 REFERENCES
2 Ahlberg, J.; Ahlbom, A.; Lipping, H.; Norell, S.; Osterblom, L. (1981) [Cancer among professional
3 drivers—a problem-oriented register-based study]. Lakartidningen 78: 1545-1546.
4
5 Benhamou, S.; Benhamou, E.; Flamant, R. (1988) Occupational risk factors of lung cancer in a French
6 case-control study. Br. J. Ind. Med. 45: 231-233.
7
8 Boffetta, P.; Stellman, S. D. (1988) Association between diesel exhaust exposure and multiple myeloma: an
9 example of confounding. Prev. Med. 17: 236-237.
10
11 Buiatti, E.; Kriebel, D.; Geddes, M.; Santucci, M.; Pucci, N. (1985) A case control study of lung cancer in
12 Florence, Italy. I. Occupational risk factors. J. Epidemiol. Commun. Health 39: 244-250.
13
14 Clayson, D. B.; Garner, R. C. (1976) Carcinogenic aromatic amines and related compounds. In: Searle, C. E.,
15 ed. Chemical carcinogens. Washington, DC: American Chemical Society. (ACS monograph no. 173).
16
17 Coggon, D.; Pannett, B.; Acheson, E. D. (1984) Use of job-exposure matrix in an occupational analysis of lung
18 and bladder cancers on the basis of death certificates. JNCI J. Natl. Cancer Inst. 72: 61-65.
19
20 Damber, L. A.; Larsson, L. G. (1987) Occupation and male lung cancer: a case-control study in northern
21 Sweden. Br. J. Ind. Med. 44: 446-453.
22
23 Daniel, J. H. (1984) The use of diesel-powered equipment in U.S. underground coal operations. Presented at:
24 American Mining Congress international coal show; May; Chicago, IL. Washington, DC: Department of
25 the Interior, Bureau of Mines.
26
27 Edling, C.; Anjou, C.-G.; Axelson, 0.; Kling, H. (1987) Mortality among personnel exposed to diesel exhaust.
28 Int. Arch. Occup. Environ. Health 59: 559-565.
29
30 Flodin, U.; Fredriksson, M.; Persson, B. (1987) Multiple myeloma and engine exhausts, fresh wood, and
31 creosote: a case-referent study. Am. J. Ind. Med. 12: 519-529.
32
33 Garland, F. C.; Gorham, E. D.; Garland, C. F.; Ducatman, A. M. (1988) Testicular cancer in US Navy
34 personnel. Am. J. Epidemiol. 127: 411-414.
35
36 Garshick, E.; Schenker, M. B.; Munoz, A.; Segal, M.; Smith, T. J.; Woskie, S. R.; Hammond, S. K.; Speizer,
37 F. E. (1987) A case-control study of lung cancer and diesel exhaust exposure in railroad workers.
38 Am. Rev. Respir. Dis. 135: 1242-1248.
39
40 Garshick, E.; Schenker, M. B.; Munoz, A.; Segal, M.; Smith, T. J.; Woskie, S. R.; Hammond, S. K.; Speizer,
41 F. E. (1988) A retrospective cohort study of lung cancer and diesel exhaust exposure in railroad workers.
42 Am. Rev. Respir. Dis. 137: 820-825.
43
44 Gustafsson, L.; Wall, S.; Larsson, L.-G.; Skog, B. (1986) Mortality and cancer incidence among Swedish dock
45 workers—a retrospective cohort study. Scand. J. Work Environ. Health 12: 22-26.
46
47 Gustavsson, P.; Plato, N.; Lidstrom, E.-B.; Hogstedt, C. (1990) Lung cancer and exposure to diesel exhaust
48 among bus garage workers. Scand. J. Work Environ. Health 16: 348-354.
49
50 Hall, N. E. L.; Wynder, E. L. (1984) Diesel exhaust exposure and lung cancer: a case-control study. Environ.
51 Res. 34: 77-86.
52
53 Harris, J. E. (1983) Diesel emissions and lung cancer. Risk Anal. 3: 83-100.
December 1994 g-72 DRAFT-DO NOT QUOTE OR CITE
-------
1 Hayes, R. B.; Thomas, T.; Silverman, D. T.; Vineis, P.; Blot, W. J.; Mason, T. J.; Pickle, L. W.; Correa, P.;
2 Fontham, E. T. H.; Schoenberg, J. B. (1989) Lung cancer in motor exhaust-related occupations.
3 Am. J. Ind. Med. 16: 685-695.
4
5 Hirose, M.; Lee, M.-S.; Wang, C. Y.; King, C. M. (1984) Induction of rat mammary gland tumors by
6 1-nitropyrene, a recently recognized environmental mutagen. Cancer Res. 44: 1158-1162.
7
8 Hoar, S. K.; Hoover, R. (1985) Truck driving and bladder cancer mortality in rural New England. JNCI J. Natl.
9 Cancer Inst. 74: 771-774.
10
11 Howe, G. R.; Burch, J. D.; Miller, A. B.; Cook, G. M.; Esteve, J.; Morrison, B.; Gordon, P.; Chambers,
12 L. W.; Fodor, G.; Winsor, G. M. (1980) Tobacco use, occupation, coffee, various nutrients, and bladder
13 cancer. JNCI J. Natl. Cancer Inst. 64: 701-713.
14
15 Howe, G. R.; Fraser, D.; Lindsay, J.; Presnal, B.; Yu, S. Z. (1983) Cancer mortality (1965-77) in relation to
16 diesel fume and coal exposure in a cohort of retired railway workers. JNCI J. Natl. Cancer Inst.
17 70: 1015-1019.
18
19 Humble, C. G.; Samet, J. M.; Skipper, B. E. (1984) Comparison of self- and surrogate-reported dietary
20 information. Am. J. Epidemiol. 119: 86-98.
21
22 Iscovich, J.; Castelleto, R.; Esteve, J.; Mufloz, N.; Colanzi, R.; Coronel, A.; Deamezola, L; Tassi, V.;
23 Arslan, A. (1987) Tobacco smoking, occupational exposure and bladder cancer in Argentina. Int.
24 J. Cancer 40: 734-740.
25
26 Iyer, V.; Harris, R. E.; Wynder, E. L. (1990) Diesel exhaust exposure and bladder cancer risk. Eur.
27 J. Epidemiol. 6: 49-54.
28
29 Jensen, O. M.; Wahrendorf, J.; Knudsen, J. B.; S0rensen, B. L. (1987) The Copenhagen case-referent study on
30 bladder cancer: risks among drivers, painters, and certain other occupations. Scand. J. Work Environ.
31 Health 13: 129-134.
32
33 Kaplan, I. (1959) Relationship of noxious gases to carcinoma of the lung in railroad workers. JAMA
34 J. Am. Med. Assoc. 171: 2039-2043.
35
36 Kolonel, L. N.; Hirohata, T.; Nomura, A. M. Y. (1977) Adequacy of survey data collected from substitute
37 respondents. Am. J. Epidemiol. 106: 476-484.
38
39 Lerchen, M. L.; Samet, J. M. (1986) An assessment of the validity of questionnaire responses provided by a
40 surviving spouse. Am. J. Epidemiol. 123: 481-489.
41
42 Lerchen, M. L.; Wiggins, C. L.; Samet, J. M. (1987) Lung cancer and occupation in New Mexico. JNCI
43 J. Natl. Cancer Inst. 79: 639-645.
44
45 McLaughlin, J. K.; Dietz, M. S.; Mehl, E. S.; Blot, W. J. (1987) Reliability of surrogate information on
46 cigarette smoking by type of informant. Am. J. Epidemiol. 126: 144-146.
47
48 Percy, C.; Stanek, E., Ill; Gloeckler, L. (1981) Accuracy of cancer death certificates and its effect on cancer
49 mortality statistics. Am. J. Public Health 71: 242-250.
50
51 Raffle, P. A. B. (1957) The health of the worker. Br. J. Ind. Med. 14: 73-80.
52
53 Risen, H. A.; Burch, J. D.; Miller, A. B.; Hill, G. B.; Steele, R.; Howe, G. R. (1988) Occupational factors and
54 the incidence of cancer of the bladder in Canada. Br. J. Ind. Med. 45: 361-367.
December 1994 8-73 DRAFT-DO NOT QUOTE OR CITE
-------
1 Rogot, E.; Reid, D. D. (1975) The validity of data from next-of-kin in studies of mortality among migrants. Int.
2 J. Epidemiol. 4: 51-54.
3
4 Rosenblatt, M. B.; Teng, P. K.; Kerpe, S.; Beck, I. (1971) Causes of death in 1,000 consecutive autopsies.
5 N. Y. State J. Med. 71: 2189-2193.
6
7 Rosenkranz, H. S.; Mermelstein, R. (1983) Mutagenicity and genotoxicity of nitroarenes: all nitro-containing
8 chemicals were not created equal. Mutat. Res. 114: 217-267.
9
10 Rushton, L.; Alderson, M. R.; Nagarajah, C. R. (1983) Epidemiological survey of maintenance workers in
11 London Transport Executive bus garages at Chiswick Works. Br. J. Ind. Med. 40: 340-345.
12
13 Schenker, M. B.; Smith, T.; Munoz, A.; Woskie, S.; Speizer, F. E. (1984) Diesel exposure and mortality
14 among railway workers: results of a pilot study. Br. J. Ind. Med. 41: 320-327.
15
16 Schottenfeld, D.; Eaton, M.; Sommers, S. C.; Alonso, D. R.; Wilkinson, C. (1982) The autopsy as a measure
17 of accuracy of the death certificate. Bull. N. Y. Acad. Med. 58: 778-794.
18
19 Siemiatycki, J.; Gerin, M.; Stewart, P.; Nadon, L.; Dewar, R.; Richardson, L. (1988) Associations between
20 several sites of cancer and ten types of exhaust and combustion products: results from a case-referent
21 study in Montreal. Scand. J. Work Environ. Health 14: 79-90.
22
23 Silverman, D. T.; Hoover, R. N.; Albert, S.; Graff, K. M. (1983) Occupation and cancer of the lower urinary
24 tract in Detroit. JNCI J. Natl. Cancer Inst. 70: 237-245.
25
26 Silverman, D. T.; Hoover, R. N.; Mason, T. J.; Swanson, G. M. (1986) Motor exhaust-related occupations and
27 bladder cancer. Cancer Res. 46: 2113-2116.
28
29 Steenland, K. (1986) Lung cancer and diesel exhaust: a review. Am. J. Ind. Med. 10: 177-189.
30
31 Steenland, K.; Burnett, C.; Osoria, A. M. (1987) A case-control study of bladder cancer using city directories as
32 a source of occupational data. Am. J. Epidemiol. 126: 247-257.
33
34 Steenland, N. K.; Silverman, D. T.; Hornung, R. W. (1990) Case-control study of lung cancer and truck driving
35 in the Teamsters Union. Am. J. Public Health 80: 670-674.
36
37 Steineck, G.; Plato, N.; Gerhardsson, M.; Norell, S. E.; Hogstedt, C. (1990) Increased risk of urothelial cancer
38 in Stockholm during 1985-87 after exposure to benzene and exhausts. Int. J. Cancer 45: 1012-1017.
39
40 Stenberg, U.; Alsberg, T.; Westerholm, R. (1983) Emission of carcinogenic components with automobile
41 exhausts. Environ. Health Perspect. 47: 53-63.
42
43 Stern, F. B.; Lemen, R. A.; Curtis, R. A. (1981) Exposure of motor vehicle examiners to carbon monoxide:
44 a historical prospective mortality study. Arch. Environ. Health 36: 59-66.
45
46 Vineis, P.; Magnani, C. (1985) Occupation and bladder cancer in males: a case-control study. Int. J. Cancer
47 35: 599-606.
48
49 Vostal, J. J. (1986) Factors limiting the evidence for chemical carcinogenicity of diesel emissions in long-term
50 inhalation experiments. In: Ishinishi, N.; Koizumi, A.; McClellan, R. O.; Stober, W., eds. Carcinogenic
51 and mutagenic effects of diesel engine exhaust: proceedings of the international satellite symposium on
52 lexicological effects of emissions from diesel engines; July; Tsukuba Science City, Japan. Amsterdam,
53 The Netherlands: Elsevier Science Publishers B. V.; pp. 381-396. (Developments in toxicology and
54 environmental science: v. 13).
December 1994 8-74 DRAFT-DO NOT QUOTE OR CITE
-------
1 Waller, R. E. (1981) Trends in lung cancer in London in relation to exposure to diesel fumes. Environ. Int.
2 5: 479-483.
3
4 Waxweiler, R. J.; Wagoner, J. K.; Archer, V. E. (1973) Mortality of potash workers. J. Occup. Med.
5 15:486-489.
6
7 Williams, R. R.; Stegens, N. L.; Goldsmith, J. R. (1977) Associations of cancer site and type with occupation
8 and industry from the Third National Cancer Survey interview. J. Natl. Cancer Inst. 59: 1147-1185.
9
10 Wong, O.; Morgan, R. W.; Kheifets, L.; Larson, S. R.; Whorton, M. D. (1985) Mortality among members of a
11 heavy construction equipment operators union with potential exposure to diesel exhaust emissions.
12 Br. J. Ind. Med. 42: 435-448.
13
14 Woskie, S. R.; Smith, T. J.; Hammond, S. K.; Schenker, M. B.; Garshick, E.; Speizer, F. E. (1988a)
15 Estimation of the diesel exhaust exposures of railroad workers: I. current exposures. Am. J. Ind. Med.
16 13: 381-394.
17
18 Woskie, S. R.; Smith, T. J.; Hammond, S. K.; Schenker, M. B.; Garschick, E.; Speizer, F. E. (1988b)
19 Estimation of the diesel exhaust exposures of railroad workers: II. national and historical exposures.
20 Am. J. Ind. Med. 13: 395-404.
21
22 Wynder, L. E.; Dieck, G.; Hall, N. E. (1985) A case control study of diesel exhaust exposure and bladder
23 cancer. Environ. Res. 34: 475-489.
December 1994 8-75 DRAFT-DO NOT QUOTE OR CITE
-------
i ADDENDUM TO CHAPTER 8
2
3 Since the last update of this document, eight epidemiologic studies have been published.
4 Of these studies, four were excluded from this review. Three studies (Swanson et al., 1993;
5 Cordier et al., 1993; and Notani et al., 1993) were hypothesis-generating studies and would
6 not have contributed anything towards the evaluation of the carcinogenicity of diesel exhaust.
7 The fourth study (Guberan et al., 1992), which failed to distinguish between diesel exhaust
8 and gasoline exhaust, was excluded due to uncertain contribution of diesel exhaust to the
9 observed increase for lung cancer.
10 The review of the remaining four studies appears in this addendum.
11
12 Gustavsson et al. (1990): Lung Cancer and Exposure to Diesel Exhaust
13 Among Bus Garage Workers
14 A retrospective mortality study (from 1952 to 1986), cancer incidence study (from 1958
15 to 1984), and nested case-control study were conducted among a cohort of 708 male workers
16 from five bus garages in Stockholm, Sweden, who had worked for at least 6 mo between
17 1945 and 1970. Thirteen individuals were lost to follow-up, reducing the cohort to 695.
18 Information was available on location of workplace, job type, and beginning and ending
19 of work periods. Workers were traced using a computerized register of the living
20 population, death and burial books, and data from the Stockholm City archives.
21 For the cohort mortality analyses, death rates of the general population of greater
22 Stockholm were used. Death rates of occupationally active individuals, a subset of the
23 general population of greater Stockholm, were used as a second comparison group to reduce
24 the bias from "healthy worker effect". Mortality analysis was conducted using the
25 "occupational mortality analysis program" (OCMAP-PC). For cancer incidence analysis, the
26 "epidemiology in Linkoping" (EPILIN) program was utilized using the incidence rates
27 obtained from the cancer registry.
28 For the nested case-control study, both dead and incident primary lung cancers,
29 identified in the register of cause of deaths and the cancer register, were selected as cases
30 (20). Six controls matched on age ±2 years, selected from the noncases at the time of the
31 diagnosis of cases, were drawn at random without replacements. Matched analyses were
December 1994 8A4 DRAFT-DO NOT QUOTE OR CITE
-------
1 done to calculate odds ratios using conditional logistic regression. The EGRET and Epilog
2 programs were used for these analyses.
3 Diesel exhaust and asbestos exposure assessments were performed by industrial
4 hygienists based on the intensity of exposure to diesel exhaust and asbestos, specific for
5 workplace, work task, and calendar time period. A diesel exhaust exposure assessment was
6 based on (1) amount of emission (number of buses, engine size, running time, and type of
7 fuel), (2) ventilatory equipment and air volume of the garages, and (3) job types and work
8 practices. Based on detailed historical data and very few actual measurements, relative
9 exposures were estimated (these were not absolute exposure levels). The scale was set to
10 0 for unexposed and 1 for lowest exposure with each additional unit increase corresponding
11 to a 50% increase in successive intensity (i.e., 1.5, 2.25, 3.38, and 5.06).
12 Based on personal sampling of asbestos during 1987, exposures were estimated and
13 time-weighted annual mean exposures were classified on a scale of three degrees (0,1,
14 and 2). Cumulative exposures for both diesel exhaust and asbestos were calculated by
15 multiplying the level of exposure by the duration of every work period. An exposure index
16 was calculated by adding for every individual contributions from all work periods for both
17 diesel exhaust and asbestos. Four diesel exhaust index classes were created: (1) 0 to 10,
18 (2) 10 to 20, (3) 20 to 30, and (4) >30. The four asbestos index classes were (1) 0 to 20,
19 (2) 20 to 40, (3) 40 to 60, and (4) >60. The cumulative exposure indices were used for the
20 nested case-control study.
21 Excesses were observed for all cancers and some other site-specific cancers using both
22 comparison populations for the cohort mortality study—but none of them was statistically
23 significant. Based on 17 cases, standardized mortality ratios (SMRs) for lung cancer were
24 122 and 115 using Stockholm occupationally active and general population, respectively.
25 No dose-response was observed with increasing cumulative exposure. The cancer incidence
26 study reportedly confirmed the mortality results (results not given).
27 The nested case-control study showed increasing risk of lung cancer with increasing
28 exposure. Weighted linear regression gave RRs of 1.34 (95% CI = 1.09 to 1.64),
29 1.81 (95% CI = 1.20 to 2.71), and 2.43 (95% CI = 1.32 to 4.47) for the diesel exhaust
30 indices 10 to 20, 20 to 30, and > 30, respectively, using 0 to 10 as the comparison group.
31 The results from conditional logistic regression were similar to those obtained by weighted
December 1994 8A-2 DRAFT-DO NOT QUOTE OR CITE
-------
1 linear regression but none was statistically significant. Adjustment for asbestos exposure did
2 not change the lung cancer risk for diesel exhaust.
3 The main strength of this study is the detailed exposure matrices constructed for both
4 diesel exhaust and asbestos exposure, although they were based primarily on job tasks and
5 very few actual measurements. There are a few methodological limitations to this study.
6 The cohort is small and there were only 17 lung cancer deaths, thus the power is low.
7 Exposure or outcome may be misclassified, although any resulting bias in the relative risk
8 estimates is likely to be towards unity, because exposure classification was done
9 independently of the outcome. Although the analysis by dose indices was done, no latency
10 analysis was performed. Finally, data on smoking were missing, thus potentially
11 confounding the lung cancer results. The authors suggest that even the heaviest smoking
12 among individuals who were heavily exposed to diesel exhaust will be unable to explain the
13 excess relative risk of 2.4 observed in this group. This may be an overstatement, however,
14 as cigarette smoking is a very strong risk factor for lung cancer. Overall, this study provides
15 some support to the excess lung cancer results found earlier among populations exposed to
16 diesel exhaust.
17
18 Boffetta et al. (1990): Case-Control Study on Occupational Exposure to
19 Diesel Exhaust and Lung Cancer Risk
20 This is an ongoing (since 1969) case-control study of tobacco-related diseases in
21 18 hospitals (six U.S. cities). Cases comprise 2,584 males, with histologically confirmed
22 primary lung cancers. Sixty-nine cases were matched to one control, whereas 2,515 were
23 matched to two controls. Controls were individuals who were diagnosed with non-tobacco-
24 related diseases. The matching was done for sex, age (±2 years), hospital, and year of
25 interview. The interviews were conducted at the hospitals at the tune of diagnosis. In .1985,
26 the occupational section of the questionnaire was modified to include the usual occupation
27 and up to five other jobs as well as duration (in years) worked in those jobs. After 1985,
28 information was also obtained on exposure to 45 groups of chemicals including diesel exhaust
29 at the workplace or during hobby activities. A priori aggregation of occupations was
30 categorized into low probability of diesel exhaust exposure (reference group), possible
31 exposure (19 occupations), and probable exposure (13 occupations). Analysis was conducted
December 1994 8A_3 DRAFT-DO NOT QUOTE OR CITE
-------
1 based on "usual occupation" on all study subjects and any occupation with sufficient cases
2 was eligible for further analysis. In addition, cases enrolled after 1985 for which there were
3 self-reported diesel exhaust exposure and detailed work histories were also analyzed
4 separately.
5 Both matched and unmatched analyses were done by calculating the adjusted (for
6 smoking and education) relative odds using the Mantel-Haenzael method and calculating the
7 test-based 95% confidence interval using the Miettinen method. Unconditional logistic
8 regression was used to adjust for potential confounders (the PROC LOGIST of SAS). Linear
9 trends for risk were also tested according to Mantel.
10 Adjusted relative odds for possible and probable exposure groups as well as the truck
11 drivers were slightly below unity, none being statistically significant for the entire study
12 population. Although slight excesses were observed for the self-reported diesel exhaust
13 exposure group and the subset of post-1985 enrollees for highest duration of exposure (for
14 self-reported exposure, occupations with probable exposure and for truck drivers), none was
15 statistically significant. Trend tests for the risk of lung cancer among self-reported diesel
16 exhaust exposure, probable exposure, and truck drivers with increasing exposure (duration of
17 exposure used as surrogate for increasing dose) were nonsignificant too. Statistically
18 significant lung cancer excesses were observed for cigarette smoking only.
19 The major strength of this study is availability of detailed smoking history. Even
20 though detailed information was obtained for the usual and five other occupations (1985), no
21 effort was made to estimate or verify the actual exposure to diesel exhaust; instead, duration
22 of employment was used as a surrogate for dose. The numbers of cases and controls were
23 large; however, the number of individuals exposed to diesel exhaust was relatively few, thus
24 reducing the power of the study. This study did not attempt latency analysis either. Given
25 these limitations, the findings of this study are unable to provide either positive or negative
26 evidence for a causal association between diesel exhaust and occurrence of lung cancer.
27
28 Hansen (1993): A Follow-up Study on the Mortality of Truck Drivers
29 This is a retrospective cohort mortality study of unskilled male laborers, ages 15 to
30 74 years, in Denmark, identified from a nationwide census file of November 9, 1970. The
31 exposed group included all truck drivers employed in the road delivery or long-haul business
December 1994 8A-4 DRAFT-DO NOT QUOTE OR CITE
-------
1 (14,225). The unexposed group included all laborers in certain selected occupational groups
2 considered to be unexposed to fossil fuel combustion products and to resemble truck drivers
3 in terms of work-related physical demands and various personal background characteristics
4 (43,024).
5 Through automatic record linkage between the 1970 census register (the Central
6 Population Register 1970 to 1980) and the Death Certificate Register (1970 to 1980), the
7 population was followed for cause-specific mortality or emigration up to November 9, 1980.
8 Expected number of deaths among truck drivers were calculated by using the 5-year age
9 group and 5-year time period death rates of the unexposed group and applying them to the
10 person-years accumulated by truck drivers. International Classification of Diseases
11 Revision 8 was used to code the underlying cause of death. Test based confidence intervals
12 (CI) were calculated using Miettinen's Method. A Poisson distribution was assumed for the
13 smaller numbers and CI were calculated based on exact Poisson distribution (Ciba-Geigy).
14 Total person-years accrued by truck drivers were 138,302, whereas, for the unexposed
15 population, they were 407,780. There were 627 deaths among truck drivers and
16 3,811 deaths in the unexposed group. Statistically significant (SS) excesses were observed
17 for all cancer mortality (SMR = 121, 95% CI = 104 to 140); cancer of respiratory organs
18 (SMR = 160, 95% CI = 128 to 198), which mainly was due to cancer of bronchus and lung
19 (SMR = 160, 95% CI = 126 to 200); and multiple myeloma (SMR = 439, 95% CI = 142
20 to 1,024). When lung cancer mortality was further explored by age groups, excesses were
21 observed in most of the age groups (30 to 39, 45 to 99, 50 to 54, 55 to 59, 60 to 64, and
22 65 to 74), but there were small numbers of deaths in each group when stratified by age, and
23 the excesses were statistically significant for the 55 to 59 (SMR = 229, 95% CI = 138 to
24 358) and 60 to 64 (SMR = 227, 95% CI = 142 to 344) age groups only.
25 As acknowledged by the author, the study has quite a few methodologic limitations.
26 The exposure to diesel exhaust is assumed in truck drivers based on diesel-powered trucks,
27 but no validation of qualitative or quantitative exposure is attempted. It is also not known
28 whether any of these truck drivers or any other laborers had changed jobs after the census of
29 November 9, 1970, thus creating potential misclassification bias in exposure to diesel
30 exhaust. The lack of smoking data and a 36% rural population (usually consuming less
31 tobacco) in the unexposed group further confound the lung cancer results. The follow-up
December 1994 8A_5 DRAFT-DO NOT QUOTE OR CITE
-------
1 period is relatively short and a latency analysis was not attempted. At best, the findings of
2 this study are consistent with the findings of other truck driver studies.
3
4 Emmelin et al. (1993): Diesel Exhaust Exposure and Smoking:
5 A Case-Referent Study of Lung Cancer Among Swedish Dock Workers
6 This is a case-control study of lung cancer drawn from the cohort defined as all male
7 workers who had been employed as dock workers for at least 6 mo between 1950 and 1974.
8 In the population of 6,573 from 20 ports, there were 90 lung cancer deaths (cases), identified
9 through Swedish death and cancer registers, during the period of 1960 to 1982. Of these
10 90 deaths, the 54 who were workers at the 15 ports for which exposure surrogate information
11 was available were chosen for the case-control study. Four controls, matched on port and
12 age, were chosen for each case from the remaining cohort who had survived to the time of
13 diagnosis of the case. Both live and deceased controls were included. The final analyses
14 were done on 50 cases and 154 controls who had complete information on employment dates
15 and smoking data. The smoking strata were created by classifying ex-smokers as
16 nonsmokers if they had not smoked for at least 5 years prior to the date of diagnosis of the
17 case; otherwise they were classified as smokers.
18 Relative odds and regression coefficients were calculated using conditional logistic
19 regression models. Comparisons were made both with and without smoking included as a
20 variable, and the possible interaction between smoking and diesel exhaust was tested. Both
21 weighted linear regressions of the adjusted relative odds, and the regression coefficients were
22 used to test mortality trends with all three exposure variables.
23 Exposure to diesel exhaust was assessed indirectly by initially measuring (1) exposure
24 intensity based on exhaust emission, (2) characteristics of the environment in terms of
25 ventilation, and (3) measures of proportion of time in higher exposed jobs. For exhaust
26 emissions annual diesel fuel consumption at a port was used as the surrogate. For ventilation
27 the annual proportion of ships with closed or semi-closed holds was used as the surrogate.
28 The proportion of time spent below decks was used as the surrogate for more exposed jobs.
29 Although data were collected for all three measures only the annual fuel consumption was
30 used for analysis. Because every man was likely to rotate through the various jobs, the
31 authors thought using annual consumption of diesel fuel was the appropriate measure of
December 1994 8A-6 DRAFT-DO NOT QUOTE OR CITE
-------
1 exposure. Consequently, in a second analysis, the annual fuel consumption was divided by
2 the number of employees in the same port that year to come up with the fuel-per-person
3 measure, which was further used to create a second measure, "exposed time". The "annual
4 fuel" and exposed-time data were entered in a calendar time-exposure matrix for each port,
5 from which individual exposure measures were created. A third measure, "machine time"
6 (years of employment from first exposure) was also used to compare the results with other
7 studies. All exposure measures were accumulated from the first year of employment or first
8 year of diesel machine use, whichever came later. The last year of exposure was fixed at
9 1979. All exposures within 2 years prior to the date of lung cancer diagnosis were omitted
10 both from cases and matched controls. A priori classification into three categories of low,
11 medium, and high exposure was done for all three exposure variables, machine time, fuel,
12 and exposed time.
13 Conditional logistic regression models, adjusting for smoking status, and using low
14 exposures and/or nonsmoker as a comparison group yielded positive trends for all exposure
15 measures, but no trend test results were reported, and only the relative odds for the
16 exposed-time exposure measure in the high-exposure group (OR = 6.8, 90% CI = 1.3 to
17 34.9), was reported as statistically significant. For smokers, adjusting for diesel exhaust
18 exposure level, the relative odds were statistically significant and about equal for all the three
19 exposure variables—machine time, OR = 5.7 (90% CI = 2.4 to 13.3); fuel, OR = 5.5
20 (90% CI = 2.4 to 12.7); and exposed time, OR = 6.2 (90% CI = 2.6 to 14.6). Interaction
21 between diesel exhaust and smoking was tested by conditional logistic regression in the
22 exposed-time variable. Although there were positive trends for both smokers and
23 nonsmokers, the trend for smokers was much steeper—low, OR = 3.7 (90% CI = 0.9 to
24 14.6); medium, OR = 10.7 (90% CI = 1.5 to 78.4); and high, OR = 28.9 (90% CI =
25 3.5 to 240) indicating more than additive interaction between these two variables.
26 In the weighted linear regression model with the exposed-time variable the results .were
27 similar to those using the logistic regression model. The authors also explored the smoking
28 variable further in various analyses, some of which suggested a strong interaction between
29 diesel exhaust and smoking. However, with just six nonsmokers and no further
30 categorization of smoking amount or duration, these results are of limited value.
31 The diesel exhaust exposure matrices created using three different variables are
December 1994 8A_7 DRAFT-DO NOT QUOTE OR CITE
-------
1 intricate. Analyses by any of these variables essentially yield the same positive results and
2 positive trends, providing consistent support for a real effect of diesel exhaust exposure, at
3 least in smokers. However, there are some methodological limitations to this study which
4 prevent a more definitive conclusion. The numbers of cases and controls are small. There
5 are very few nonsmokers, thus testing the effects of diesel exhaust exposure in them is futile.
6 Lack of information on asbestos exposure, to which dock workers are usually exposed, may
7 also confound the results. Also, no latency analyses are presented. Overall, despite these
8 limitations, this study supports the earlier findings of excess lung cancer mortality among
9 individuals exposed to diesel exhaust.
10
11 Conclusion
12 In conclusion, of these four recent studies, three provide support to the earlier findings
13 of increased lung cancer among individuals exposed to diesel exhaust, whereas one study is
14 neither able to support nor refute the earlier findings.
15
16 References
17 Boffetta, P.; Harris, R. E.; Wynder, E. L. (1990) Case-Control study on occupational exposure to diesel
18 exhaust and lung cancer risk. Am. J. Ind. Med. 17:577-591.
19
20 Cordier, S.; Clavel, J.; Limasset, J. C.; Boccon-Gibod, L.; Le Moual, N.; Mandereau, L.; Hemon, D. (1993)
21 Occupational risks of bladder cancer in France: A multicenter case-control study. Int. J. Epidemiol.
22 22:403-411.
23
24 Emmelin, A.; Nystrom, L.; Wall, S. (1993) Diesel exhaust exposure and smoking: A case-referent study of
25 lung cancer among Swedish Dock workers. Epidemiol. 4:237-244.
26
27 Guberan, E.; Usel, M.; Raymond, L.; Bolay, J.; Fioretta, G.; Puissant, J. (1992) Increased risk for lung cancer
28 and for cancer of the gastrointestinal tract among Geneva professional drivers. Br. J. Ind. Med.
29 49:337-344.
30
31 Gustavsson, P.; Plato, N.; Lindstrom, E-B.; Hogstedt, C. (1990) Lung cancer and exposure to diesel exhaust
32 among bus garage workers. Scand. J. Work. Environ. Health. 16:348-354.
33
34 Hansen, E. S. (1993) A follow-up study on the mortality of truck drivers. Am. J. Ind. Med. 23:811-821.
35
36 Notani, P. N.; Shah, P.; Jayant, K.; Balakrishnan, V. (1993) Occupation and cancers of the lung and bladder:
37 A case-control study in Bombay. Int. J. Epidemiol. 22:185-191.
38
39 Swanson, G. M.; Lin, C-S.; Burns, P. B. (1993) Diversity in the association between occupation and lung cancer
40 among black and white men. Cancer Epidemiol. Biomark Prev. 2:313-320.
December 1994 8A-8 DRAFT-DO NOT QUOTE OR CITE
-------
i 9. MUTAGENICITY
2
3
4 Since 1978, over 100 publications have appeared in which genotoxicity assays have
5 been employed with diesel emissions, the volatile and paniculate fractions (including
6 extracts), or individual chemicals found in diesel emissions. Although most of the studies
7 deal with the question of whether particulate extracts from diesel emissions possessed
8 mutagenic activity in microbial and mammalian cell assays, a number of studies in recent
9 years have employed bioassays (most commonly Salmonella TA98 without S9) to evaluate
10 (1) extraction procedures, (2) fuel modifications, (3) bioavailability of chemicals from
11 particles, and (4) exhaust filters or other modifications and other variables associated with
12 diesel emissions. This chapter will focus on the application of the available data to issues of
13 genetic risk assessment; reports dealing with mutagenic activity associated with the
14 metabolism of particular chemicals of diesel particles are discussed in Chapter 10. Also,
15 because of the large number of reports, this discussion will focus on key references. The
16 recent International Agency for Research on Cancer (IARC) monograph (International
17 Agency for Research on Cancer, 1989) contains an exhaustive description of the available
18 studies and other review articles (Claxton, 1983; Pepelko and Peraino, 1983) and the
19 proceedings of several symposia on the health effects of diesel emissions (U.S.
20 Environmental Protection Agency, 1980; Lewtas, 1982; Ishinishi et al., 1986; International
21 Agency for Research on Cancer, 1989) are also available.
22
23
24 9.1 GENE MUTATIONS
25 Huisingh et al. (1978) demonstrated that dichloromethane extracts from diesel particles
26 were mutagenic in strains TA1537, TA1538, TA98, and TAIOO of 5. typhimurium, both with
27 and without rat liver S9 activation. This first report contained data from several different
28 fractions as well as particulate material from different vehicles and different fuels. Similar
29 results with diesel extracts from various engines and fuels have been reported by a number of
30 investigators using the Salmonella frameshift sensitive strains TA1537, TA1538, and TA98
31 (Siak et al., 1981; Claxton, 1981; Dukovich et al., 1981; Brooks et al., 1984). Similarly,
December 1994 94 DRAFT-DO NOT QUOTE OR CITE
-------
1 mutagenic activity was observed in Salmonella forward mutation assays measuring
2 8-azaguanine resistance (Claxton and Kohan, 1981) and in E. coli mutation assays (Lewtas,
3 1983).
4 An approach to the identification of significant mutagens in chemically complex
5 environmental samples such as diesel exhaust or ambient paniculate extracts is the
6 combination of short-term bioassays with chemical fractionation (Scheutzle and Lewtas,
7 1986). The analysis most frequently is carried out by sequential extraction with increasingly
8 polar or binary solvents. Prefractionation is by silica-column chromatography to separate
9 compounds by polarity, or separation into acidic, basic, and neutral fractions. The resulting
10 fractions are too complex to characterize by chemical methods; however, the bioassay
11 analysis can be used to determine fractions for further analysis. In most of the applications
12 of this concept, Salmonella strain TA98 without the addition of S9 has been employed as the
13 indicator for mutagenic activity. Generally, a variety of nitrated polynuclear aromatic
14 compounds has been found, which account for a substantial portion of the mutagenicity found
15 (Liberti et al., 1984; Schuetzle and Frazer, 1986; Schuetzle and Perez,1983). However, not
16 all the bacterial mutagenicity has been identified in this way, and the identity of the
17 remainder of the mutagenic compounds remains unknown. The identity of the nitrated
18 aromatics thus far identified in diesel exhaust was the subject of review in the IARC
19 monograph on diesel exhaust (International Agency for Research on Cancer, 1989).
20 In addition to the simple qualitative identification of mutagenic chemicals, several
21 investigators have used the numerical data to express mutagenic activity as activity per
22 distance driven or mass of fuel consumed. These types of calculations have been the basis
23 for the estimates that the nitroarenes (both mono- and dinitropyrenes) contribute a significant
24 amount of the total mutagenic activity of the whole extract (Nishioka et al., 1982; Salmeen
25 et al., 1982; Nakagawa et al., 1983). However, as noted by Claxton (1983) because most of
26 these studies used only strain TA98 without exogenous activation, there are several classes of
27 mutagenic chemicals which may have gone undetected.
28 Matsushita et al. (1986) tested particle-free diesel exhaust gas and a number of benzene
29 nitro-derivatives, and poly cyclic aromatic hydrocarbons (PAHs) (many of which have been
30 identified as components of diesel exhaust gas). The particle-free exhaust gas was positive in
31 both TA100 and TA98, but only without S9 activation. Of the 94 nitro-benzene derivatives
December 1994 9_2 DRAFT-DO NOT QUOTE OR CITE
-------
1 tested, 61 were mutagenic, and the majority showed greatest activity in TA100 without S9.
2 Twenty-eight of 50 PAHs tested were mutagenic, all required the addition of S9 for
3 detection, and most appeared to show a stronger response in TA100. When
4 1,6-dinitropyrene was mixed with various PAHs or an extract of heavy-duty (HD) diesel
5 exhaust, the mutagenic activity in TA98 was greatly reduced when S9 was absent but was
6 increased significantly when S9 was present. These latter results suggested that caution
7 should be used in estimating mutagenicity (or other toxic effects) of complex mixtures from
8 the specific activity of individual components.
9 Mitchell et al. (1981) reported mutagenic activity of particle extracts of diesel emissions
10 in the mouse lymphoma L5178Y mutation assay. Positive results were seen both with and
11 without S9 activation in extracts from several different vehicles, with mutagenic activity only
12 slightly lower in the presence of S9. These findings have been confirmed in a number of
13 other mammalian cell systems using several different genetic markers. Casto et al. (1981),
14 Chescheir et al. (1981), Li and Royer (1982), and Brooks et al. (1984) all reported positive
15 responses at the HGPRT locus in CHO cells. Morimoto et al. (1986) used the APRT and
16 Ouar loci in CHO; Curren et al. (1981) used Ouar in Balb/c 3T3 cells. In all of these
17 studies, mutagenic activity was observed without S9 activation. Liber et al. (1981) used the
18 thymidine kinase (TK) locus in the TK6 human lymphoblast cell line and observed induced
19 mutagenesis only in the presence of rat liver S9 when testing a methylene chloride extract of
20 diesel exhaust. Barfnecht et al. (1982) also used the TK6 assay to identify some of the
21 chemicals responsible for this activation-dependent mutagenicity. They suggested that
22 fluoranthene, 1-methylphenanthrene, and 9-methylphenanthrene could account for over 40%
23 of the observed activity.
24 Morimoto et al. (1986) injected diesel paniculate extracts (250 to 4,000 mg/kg) into
25 pregnant Syrian hamsters and measured mutations at the APRT locus in embryo cells
26 cultivated 11 days after injection. Neutral fractions from both light-duty (LD) and HD tar
27 samples resulted in increased mutant frequency at 2,000 and 4,000 mg/kg. Belisario et al.
28 (1984) applied the Ames test to urine from Sprague-Dawley rats exposed to single
29 applications of diesel exhaust particles administered by gastric intubation, ip injection or
30 sc gelatin capsules. In all cases, dose-related increases were seen in TA98 (without and with
31 S9) from urine concentrates taken 24 h after particle administration. Urine from Swiss mice
December 1994 9.3 DRAFT-DO NOT QUOTE OR CITE
-------
1 exposed by inhalation to filtered exhaust (particle concentration 6 to 7 mg/m3) for 7 weeks
2 (Pereira et al., 198la) or Fischer 344 rats exposed to diesel exhaust particles (2 mg/m3) for
3 3 mo to 2 years was negative in Salmonella strains. Because of the large differences in
4 dosages, these findings should not be construed as conflicting.
5 Schuler and Niemeier (1981) exposed Drosophila males in a stainless steel chamber
6 connected to the 3-m3 chamber used for the chronic animal studies at EPA (see Hinners
7 et al., 1980 for details). Flies were exposed for 8 h and mated to untreated females 2 days
8 later. Althoug the frequency of sex-linked recessive lethals from treated males was not
9 different from controls, the limited sample size precluded detecting less than a threefold
10 increase over controls. The authors also suggested that flies might tolerate exposure to
11 higher concentrations for longer time periods.
12 Specific-locus mutations were not induced in (C3H X lO^Fj male mice exposed to
13 diesel exhaust 8 h/day, 7days/week for either 5 or 10 weeks (Russell et al., 1980). The
14 exhaust was a 1:18 dilution and the average particle concentration was 6 mg/m3. After
15 exposure, males were mated to T-Stock females and matings continued for the reproductive
16 life of the males. The results were unequivocally negative; no mutants were detected in
17 10,635 progeny derived from postspermatogonial cells or in 27,917 progeny derived from
18 spermatogonial cells.
19
20
21 9.2 CHROMOSOME EFFECTS
22 Mitchell et al. (1981) and Brooks et al. (1984) reported increases in sister chromatid
23 exchanges (SCE) in CHO cells exposed to paniculate extracts of emissions from both LD and
24 HD diesel engines. Morimoto et al. (1986) observed increased SCE from both LD and HD
25 diesel extracts in PHA-stimulated human lymphocyte cultures. Tucker et al. (1986) exposed
26 human peripheral lymphocyte cultures from four donors to direct diesel exhaust for up to
27 3 h. Exhaust was cooled by pumping through a plastic tube about 20 ft long; air flow was
28 1.5 L/min. Samples were taken at 16, 48, and 160 min of exposure. Cell cycle delay was
29 observed in all cultures; significantly increased SCE levels were reported for two of the four
30 cultures. Structural chromosome aberrations were induced in CHO cells by paniculate
December 1994 9.4 DRAFT-DO NOT QUOTE OR CITE
-------
1 extracts from a Nissan diesel engine (Lewtas, 1983) but not by similar extracts from an
2 Oldsmobile diesel engine (Brooks et al., 1984).
3 Pereira et al. (198la) exposed female Swiss mice to diesel exhaust 8 h/day,
4 5 days/week for 1, 3, and 7 weeks. The incidence of micronuclei and structural aberrations
5 were similar in bone marrow cells of both control and exposed mice. Increased incidence of
6 micronuclei, but not SCE, were observed in bone marrow cells of male Chinese hamsters
7 after 6 mo exposure to diesel exhaust (Pereira et al., 1981b).
8 Guerrero et al. (1981) observed a linear concentration-related increase in SCE in lung
9 cells cultured after intratracheal instillation of diesel exhaust particles at doses up to
10 20 mg/hamster. However, they did not observe any increase in SCE after 3 mo of inhalation
11 exposure to diesel exhaust particles (6 mg/m3).
12 Pereira et al. (1982) measured SCE in embryonic liver cells of Syrian hamsters.
13 Pregnant females were exposed to diesel exhaust (containing about 12 mg/m3 particles) from
14 Days 5 to 13 of gestation or injected intraperitoneally with diesel particles or particle extracts
15 on Gestational Day 13 (18 h before sacrifice). Neither the incidence of SCE nor mitotic
16 index were affected by exposure to diesel exhaust. The injection of particle extracts, but not
17 diesel particles resulted in a dose-related increase in SCE; however, the toxicity of the
18 particles was about twofold greater than the diesel extract.
19 In the only studies with mammalian germ cells, Russell et al. (1980) reported no
20 increase in either dominant lethals or heritable translocations in males of T-stock mice
21 exposed by inhalation to diesel emissions. In the dominant lethal test, T-Stock males were
22 exposed for 7.5 weeks and immediately mated to females of different genetic backgrounds
23 (T-stock; [C3H X 101]; [C3H X C57BL/6]; [SEC X C57BL/6]). There were no
24 differences from controls in any of the parameters measured in this assay. For heritable
25 translocation analysis, T-stock males were exposed for 4.5 weeks, mated to (SEC x
26 C57BL/6) females and the Pl males were tested for the presence of heritable translocations.
27 Although no translocations were detected among 358 progeny tested, the historical control
28 incidence is less than 1/1,000.
29
30
December 1994 9.5 DRAFT-DO NOT QUOTE OR CITE
-------
1 9.3 OTHER GENOTOXIC EFFECTS
2 Pereira et al. (1981b) exposed males of strain A mice to diesel exhaust emissions for
3 31 or 39 weeks using the same exposure regimen as noted in the previous section. Analyses
4 of caudal sperm for sperm-head abnormalities was conducted independently in three separate
5 laboratories. Although the incidence of sperm abnormalities was not significantly above
6 controls in any of the three laboratories, there were extremely large differences in scoring
7 among the three (control values were 9.2, 14.9, and 27.8% in the three laboratories).
8 Conversely, male Chinese hamsters exposed for 6 mo (Pereira et al., 1981c) exhibited almost
9 a threefold increase in sperm-head abnormalities. It is noted that the control incidence in the
10 Chinese hamsters was less than 0.5%. Hence, it is not clear whether the differing responses
11 reflect true species differences or experimental artifacts.
12
13
14 9.4 SUMMARY
15 Extensive studies with Salmonella have unequivocally demonstrated mutagenic activity
16 in both particulate and gaseous fractions of diesel exhaust. In most of the studies using
17 Salmonella, diesel particle extracts and individual nitropyrenes have exhibited the strongest
18 responses in strain TA98 when no exogenous activation was provided. Gaseous fractions
19 reportedly showed greater response in TA100, whereas benzo[a]pyrene and other
20 unsubstituted PAHs are only mutagenic in the presence of S9 fractions. The induction of
21 gene mutations has been reported in several in vitro mammalian cell lines after exposure to
22 extracts of diesel particles. Note that only the TK6 human cell line did not give a positive
23 response to diesel particle extracts in the absence of S9 activation. Mutagenic activity was
24 recovered in urine from animals treated with diesel particulate by gastric intubation, ip and
25 sc implants but not by inhalation of diesel particles or diluted diesel exhaust. Dilutions of
26 whole diesel exhaust did not induce sex-linked recessive lethals in Drosophila or specific-
27 locus mutations in male mouse germ cells.
28 Structural chromosome aberrations and SCE in mammalian cells have been induced by
29 particles and extracts. Whole exhaust induced micronuclei but not SCE or structural
30 aberrations in bone marrow of male Chinese hamsters exposed to whole diesel emissions for
31 6 mo. In a shorter exposure (7 weeks), neither micronuclei nor structural aberrations were
December 1994 9.5 DRAFT-DO NOT QUOTE OR CITE
-------
1 increased in bone marrow of female Swiss mice. Likewise, whole diesel exhaust did not
2 induce dominant lethals or heritable translocations in male mice exposed for 7.5 and
3 4.5 weeks, respectively.
4 Mutagenicity data have been applied both to issues of heritable genetic risk and somatic
5 cell effects—most notably cancer. For heritable genetic effects, the U.S. Environmental
6 Protection Agency's Guidelines for Mutagenicity Risk Assessment (Federal Register, 1986)
7 are applicable here. The mammalian germ cell studies measuring defined genetic endpoints
8 conducted on diesel emissions have shown negative results; however, the sample size in the
9 heritable translocation test is too small for a meaningful conclusion. In the absence of
10 definitive mammalian germ-cell results, the guidelines recommend that mutagenic activity
11 and the ability to interact with mammalian germ cells be evaluated separately. As stated, the
12 presence of a large number of mutagenic chemicals in diesel emissions is unambiguous.
13 Sperm abnormality assays are presumably the only other source of data on the interaction of
14 diesel emissions with mammalian germ cells. The negative response in the mouse is in
15 apparent conflict with the positive observation in the hamster and there is not sufficient
16 information to resolve this discrepancy. Hence, the questions of germ-cell interaction and
17 the potential for human germ-cell mutagenic risk of diesel emissions remain unanswered.
18 The application of genotoxicity information to the question of the potential
19 carcinogenicity of chemical agents was initially based on the premise that somatic mutation is
20 an integral step in the carcinogenic process. However, unlike the situation for germ cell
21 mutagenicity, assays are not weighted strictly by their biological relationship to the particular
22 species, sex, or tissue site of cancer. The size of the data base and the degree of correlation
23 of genotoxicity test results with animal cancer bioassays are frequently given great weight.
24 Indeed, a common conclusion of the efforts of the National Toxicology Program on the use
25 of in vitro assays is that no single in vitro genotoxicity test or battery of tests (among the
26 four assays in their program) improve on the performance of the Salmonella assay in
27 predicting rodent carcinogenicity of an untested chemical. When rodent carcinogenicity data
28 are available, phylogenetic and other biological aspects of the genotoxicity data are important
29 considerations in the weight-of-evidence process. With diesel emissions, additional
30 complications arise because of the chemical complexity of the material being tested.
31 Although it is clear that several of the individual chemical constituents of diesel exhaust have
December 1994 9.7 DRAFT-DO NOT QUOTE OR CITE
-------
1 been demonstrated to be both mutagenic and carcinogenic, it is likely that the constituents
2 responsible for the mutational increases observed in bacteria are different from those
3 responsible for the observed increases in CHO cells (Li and Dutcher, 1983) or in human
4 hepatoma-derived cells (Eddy et al., 1986). Chapter 10 deals more thoroughly with
5 metabolism and mechanisms of carcinogenesis.
6
7
December 1994 9_8 DRAFT-DO NOT QUOTE OR CITE
-------
1 REFERENCES
2 Barfknecht, T. R.; Hites, R. A.; Cavaliers, E. L.; Thilly, W. G. (1982) Human cell mutagenicity of polycyclic
3 aromatic hydrocarbon components of diesel emissions. In: Lewtas, J., ed. Toxicological effects of
4 emissions from diesel engines: proceedings of the Environmental Protection Agency 1981 diesel
5 emissions symposium; October 1981; Raleigh, NC. New York, NY: Elsevier Biomedical; pp. 277-294.
6 (Developments on toxicology and environmental science: v. 10).
7
8 Belisario, M. A.; Buonocore, V.; De Marinis, E.; De Lorenzo, F. (1984) Biological availability of mutagenic
9 compounds adsorbed onto diesel exhaust paniculate. Mutat. Res. 135: 1-9.
10
11 Brooks, A. L.; Li, A. P.; Dutcher, J. S.; Clark, C. R.; Rothenberg, S. J.; Kiyoura, R.; Bechtold, W. E.;
12 McClellan, R. O. (1984) A comparison of genotoxicity of automobile exhaust particles from laboratory
13 and environmental sources. Environ. Mutagen. 6: 651-668.
14
15 Casto, B. C.; Hatch, G. G.; Huang, S. L.; Huisingh, J. L.; Nesnow, S.; Waters, M. D. (1981) Mutagenic and
16 carcinogenic potency of extracts of diesel and related environmental emissions: in vitro mutagenesis and
17 oncogenic transformation. Environ. Int. 5: 403-409.
18
19 Claxton, L. D. (1981) Mutagenic and carcinogenic potency of diesel and related environmental emissions:
20 Salmonella bioassay. Environ. Int. 5: 389-391.
21
22 Claxton, L. D. (1983) Characterization of automotive emissions by bacterial mutagenesis bioassay: a review.
23 Environ. Mutagen. 5: 609-631.
24
25 Claxton, L.; Kohan, M. (1981) Bacterial mutagenesis and the evaluation of mobile-source emissions. In: Waters,
26 M. D.; Sandhu, S. S.; Huisingh, J. L.; Claxton, L.; Nesnow, S., eds. Short-term bioassays in the
27 analysis of complex environmental mixtures II: proceedings of the second symposium on the application
28 of short-term bioassays in the fractionation and analysis of complex environmental mixtures; March 1980;
29 Williamsburg, VA. New York, NY: Plenum Press; pp. 299-317. (Hollaender, A.; Welch, B. L.;
30 Probstein, R. F., eds. Environmental science research series: v. 22).
31
32 Curren, R. D.; Kouri, R. E.; Kim, D. M.; Schectman, L. M. (1981) Mutagenic and carcinogenic potency of
33 extracts from diesel related environmental emissions: simultaneous morphological transformation and
34 mutagenesis in BALB/c 3T3 cells. Environ. Int. 5: 411-415.
35
36 Dukovich, M.; Yasbin, R. E.; Lestz, S. S.; Risby, T. H.; Zweidinger, R. B. (1981) The mutagenic and
37 SOS-inducing potential of the soluble organic fraction collected from diesel paniculate emissions.
38 Environ. Mutagen. 3: 253-264.
39
40 Eddy, E. P.; McCoy, E. C.; Rosenkranz, H. S.; Mermelstein, R. (1986) Dichotomy in the mutagenicity and
41 genotoxicity of nitropyrenes: apparent effect of the number of electrons involved in nitroreduction Mutat
42 Res. 161: 109-111.
43
44 Federal Register. (1986) Guidelines for mutagenicity risk assessment. F. R. (September 24) 51: 34006-34012.
45
46 Guererro, R. R.; Rounds, D. E.; Orthoefer, J. (1981) Sister chromatid exchange analysis of Syrian hamster lung
47 cells treated in vivo with diesel exhaust particulates. Environ. Int. 5: 445-454.
49 Hinners, R. G.; Burkart, J. K.; Malanchuk, M. (1980) Facilities for diesel exhaust studies. In: Pepelko, W. E.;
50 Danner, R. M.; Clarke, N. A., eds. Health effects of diesel engine emissions: proceedings of an
51 international symposium; December 1979. Cincinnati, OH: U.S. Environmental Protection Agency,
52 Health Effects Research Laboratory; pp. 681-697; EPA report no. EPA-600/9-80-057b. Available from-
53 NTIS, Springfield, VA; PB81-173817.
December 1994 9.9 DRAFT-DO NOT QUOTE OR CITE
-------
1 Huisingh, J.; Bradow, R.; lungers, R.; Claxton, L.; Zweidinger, R.; Tejada, S.; Bumgarner, J.; Duffield, F.;
2 Waters, M.; Simmon, V. F.; Hare, C.; Rodriguez, C.; Snow, L. (1978) Application of bioassay to the
3 characterization of diesel particle emissions. In: Waters, M. D.; Nesnow, S.; Huisingh, J. L.; Sandhu,
4 S. S.; Claxton, L., eds. Application of short-term bioassays in the fractionation and analysis of complex
5 environmental mixtures: [proceedings of a symposium; February; Williamsburg, VA]. New York, NY:
6 Plenum Press; pp. 383-418. (Hollaender, A.; Probstein, F.; Welch, B. L., eds. Environmental science
7 research: v. 15).
8
9 International Agency for Research on Cancer. (1989) Diesel and gasoline engine exhausts and some nitroarenes.
10 Lyon, France: World Health Organization; pp. 41-185. (IARC monographs on the evaluation of
11 carcinogenic risks to humans: v. 46).
12
13 Ishinishi, N.; Koizumi, A.; McClellan, R. O.; Stober, W., eds. (1986) Carcinogenic and mutagenic effects of
14 diesel engine exhaust: proceedings of the international satellite symposium on lexicological effects of
15 emissions from diesel engines; July; Tsukuba Science City, Japan. Amsterdam, The Netherlands: Elsevier
16 Science Publishers B. V. (Developments in toxicology and environmental science: v. 13).
17
18 Lewtas, J. (1982) Mutagenic activity of diesel emissions. In: Lewtas, J., ed. lexicological effects of emissions
19 from diesel engines: proceedings of the Environmental Protection Agency 1981 diesel emissions
20 symposium; October 1981; Raleigh, NC. New York, NY: Elsevier Biomedical; pp. 243-264.
21 (Developments in toxicology and environmental science: v. 10).
22
23 Lewtas, J. (1983) Evaluation of the mutagenicity and carcinogenicity of motor vehicle emissions in short-term
24 bioassays. Environ. Health Perspect. 47: 141-152.
25
26 Li, A. P.; Dutcher, J. S. (1983) Mutagenicity of mono-, di-, and tri-nitropyrenes in Chinese hamster ovary cells.
27 Mutat. Res. 119: 387-392.
28
29 Li, A. P.; Royer, R. E. (1982) Diesel-exhaust-particle extract enhancement of chemical-induced mutagenesis in
30 cultured Chinese hamster ovary cells: possible interaction of diesel exhaust with environmental chemicals.
31 Mutat. Res. 103: 349-355.
32
33 Liber, H. L.; Andon, B. M.; Kites, R. A.; Thilly, W. G. (1981) Diesel soot: mutation measurements in bacterial
34 and human cells. Environ. Int. 5: 281-284.
35
36 Liberti, A.; Ciccioli, P.; Cecinato, A.; Brancaleoni, E.; Di Palo, C. (1984) Determination of
37 nitrated-polyaromatic hydrocarbons (nitro-PAHs) in environmental samples by high resolution
38 chromatographic techniques. J. High Resolut. Chromatogr. Chromatogr. Commun. 7: 389-397.
39
40 Matsushita, H.; Goto, S.; Endo, O.; Lee, J.-H.; Kawai, A. (1986) Mutagenicity of diesel exhaust and related
41 chemicals. In: Ishinishi, N.; Koizumi, A.; McClellan, R. O.; Stober, W., eds. Carcinogenic and
42 mutagenic effects of diesel engine exhaust: proceedings of the international satellite symposium on
43 lexicological effects of emissions from diesel engines; July; Tsukuba Science Cily, Japan. Amsterdam,
44 The Netherlands: Elsevier Science Publishers B. V.; pp. 103-118. (Developments on toxicology and
45 environmental science: v. 13).
46
47 Mitchell, A. D.; Evans, E. L.; Jolz, M. M.; Riccio, E. S.; Mortelmans, K. E.; Simmon, V. F. (1981)
48 Mulagenic and carcinogenic potency of exlracts of diesel and related environmenlal emissions: in vilro
49 mutagenesis and DNA damage. Environ. Int. 5: 393-401.
50
51
December 1994 9-10 DRAFT-DO NOT QUOTE OR CITE
-------
1 Morimoto, K.; Kitamura, M.; Kondo, H.; Koizumi, A. (1986) Genotoxicity of diesel exhaust emissions in a
2 battery of in-vitro short-term bioassays. In: Ishinishi, N.; Koizumi, A.; McClellan, R. O.; Stober, W.,
3 eds. Carcinogenic and mutagenic effects of diesel engine exhaust: proceedings of the international satellite
4 symposium on lexicological effects of emissions from diesel engines; July; Tsukuba Science City, Japan.
5 Amsterdam, The Netherlands: Elsevier Science Publishers B. V.; pp. 85-102. (Developments in
6 toxicology and environmental science: v. 13).
7
8 Nakagawa, R.; Kitamori, S.; Horikawa, K.; Nakashima, K.; Tokiwa, H. (1983) Identification of dinitropyrenes
9 in diesel-exhaust particles: their probable presence as the major mutagens. Mutat. Res. 124: 201-211.
10
11 Nishioka, M. G.; Petersen, B. A.; Lewtas, J. (1982) Comparison of nitro-aromatic content and direct-acting
12 mutagenicity of diesel emissions. In: Cooke, M.; Dennis, A. J.; Fisher, G. L., eds. Polynuclear aromatic
13 hydrocarbons: physical and biological chemistry. Columbus, OH: Battelle Press; pp. 603-613.
14
15 Pepelko, W. E.; Peirano, W. B. (1983) Health effects of exposure to diesel engine emissions: a summary of
16 animal studies conducted by the U.S. Environmental Protection Agency's Health Effects Research
17 Laboratories at Cincinnati, Ohio. J. Am. Coll. Toxicol. 2: 253-306.
18
19 Pepelko, W. E.; Danner, R. M.; Clarke, N. A., eds. (1980) Health effects of diesel engine emissions:
20 proceedings of an international symposium, v. 2; December 1979; Cincinnati, OH. Cincinnati, OH: U.S.
21 Environmental Protection Agency, Health Effects Research Laboratory; EPA report no.
22 EPA-600/9-80-057b. Available from: NTIS, Springfield, VA; PB81-173817.
23
24 Pereira, M. A.; Connor, T. H.; Meyne, J.; Legator, M. S. (1981a) Metaphase analysis, micronucleus assay and
25 urinary mutagenicity assay of mice exposed to diesel emissions. Environ. Int. 5: 435-438.
26
27 Pereira, M. A.; Sabharwal, P. S.; Gordon, L.; Wyrobek, A. J. (1981b) The effect of diesel exhaust on
28 sperm-shape abnormalities in mice. Environ. Int. 5: 459-460.
29
30 Pereira, M. A.; Sabharwal, P. S.; Kaur, P.; Ross, C. B.; Choi, A.; Dixon, T. (1981c) In vivo detection of
31 mutagenic effects of diesel exhaust by short-term mammalian bioassays. Environ. Int. 5: 439-443.
32
33 Pereira, M. A.; McMillan, L.; Kaur, P.; Gulati, D. K.; Sabharwal, P. S. (1982) Effect of diesel exhaust
34 emissions, particulates, and extract on sister chromatid exchange in transplacentally exposed fetal hamster
35 liver. Environ. Mutagen. 4: 215-220.
36
37 Russell, L. B.; Generoso, W. M.; Oakberg, E. F.; Russell, W. L.; Bangham, J. W.; Stelzner, K. F. (1980)
38 Tests for heritable effects induced by diesel exhaust in the mouse. Martin Marietta Energy Systems, Inc.,
39 Oak Ridge National Laboratory; report no. ORNL-5685.
40
41 Salmeen, L; Durisin, A. M.; Prater, T. J.; Riley, T.; Schuetzle, D. (1982) Contribution of 1-nitropyrene to
42 direct-acting Ames assay mutagenicities of diesel paniculate extracts. Mutat. Res. 104: 17-23.
43
44 Schuetzle, D.; Frazier, J. A. (1986) Factors influencing the emission of vapor and paniculate phase components
45 from diesel engines. In: Ishinishi, N.; Koizumi, A.; McClellan, R. O.; Stober, W., eds. Carcinogenic
46 and mutagenic effects of diesel engine exhaust: proceedings of the international satellite symposium on
47 lexicological effects of emissions from diesel engines; July; Tsukuba Science City, Japan. Amsterdam,
48 The Netherlands: Elsevier Science Publishers B. V.; pp. 41-63. (Developments in toxicology and
49 environmental science: v. 13).
50
51 Schuetzle, D.; Lewtas, J. (1986) Bioassay-directed chemical analysis in environmental research. Anal. Chem.
52 58: 1060A-1076A.
53
December 1994 9_H DRAFT-DO NOT QUOTE OR CITE
-------
1 Schuetzle, D.; Perez, J. M. (1983) Factors influencing the emissions of nitrated-polynuclear aromatic
2 hydrocarbons (nitro-PAH) from diesel engines. J. Air Pollut. Control Assoc. 33: 751-755.
3
4 Schuler, R. L.; Niemeier, R. W. (1981) A study of diesel emissions on Drosophila. Environ. Int. 5: 431-434.
5
6 Shelburne, J. D.; Chescheir, G. M., Ill; Garrett, N. E.; Huisingh, J. L.; Waters, M. D. (1981) Mutagenic
7 effects of environmental particulates in the CHO/HGPRT system. In: Waters, M. D.; Sandhu, S. S.;
8 Huisingh, J. L.; Claxton, L.; Nesnow, S., eds. Short-term bioassays in the analysis of complex
9 environmental mixtures II: proceedings of the second symposium on the application of short-term
10 bioassays in the fractionation and analysis of complex environmental mixtures; March 1980;
11 Williamsburg, VA. New York, NY: Plenum Press; pp. 337-350. (Hollaender, A.; Welch, B. L.;
12 Probstein, R. F., eds. Environmental science research series: v. 22).
13
14 Siak, J. S.; Chan, T. L.; Lees, P. S. (1981) Diesel particulate extracts in bacterial test systems. Environ. Int.
15 5: 243-248.
16
17 Tucker, J. D.; Xu, J.; Stewart, J.; Baciu, P. C.; Ong, T.-m. (1986) Detection of sister chromatid exchanges
18 induced by volatile genotoxicants. Teratog. Carcinog. Mutagen. 6: 15-21.
December 1994 9-12 DRAFT-DO NOT QUOTE OR CITE
-------
i 10. METABOLISM AND MECHANISM OF ACTION
2 IN DIESEL EMISSION-INDUCED CARCINOGENESIS
3
4
5 Considerable research effort has been directed toward assessing the carcinogenic
6 potential of diesel engine emissions. As indicated in Chapter 7, whole diesel exhaust is a
7 pulmonary carcinogen in rats exposed chronically to high concentrations, but the mechanism
8 lof carcinogenicity remains uncertain. The involvement of genetic or epigenetic mechanisms,
9 or a combination of these, remains to be determined regarding the pulmonary carcinogenicity
10 of diesel exhaust. That a definitive mechanism of action has eluded investigators is not
11 surprising when one considers that diesel exhaust is a complex mixture of hundreds of
12 chemicals and soot particles. In examining possible mechanisms of action of diesel exhaust-
13 induced carcinogenicity, it is necessary to address the following major areas: (1) the
14 metabolism and mechanism of action of known carcinogenic components such as
15 benzo[a]pyrene (B[a]P) and various nitroarenes, (2) the carcinogenic potential of the soot
16 particle, (3) the role of pulmonary leukocytes in the development of lung tumors, and (4) the
17 molecular dosimetry of inhaled diesel exhaust.
18
19
20 10.1 METABOLISM AND MECHANISM OF ACTION OF ORGANIC
21 CARCINOGENIC COMPONENTS OF DIESEL EXHAUST
22 The metabolism and proposed carcinogenic mechanism of action of chemicals
23 are often interrelated. It is difficult to obtain an understanding of one without consideration
24 of the other and, therefore, both of these processes are being considered in this discussion.
25 Specifically, emphasis will be given to polycyclic aromatic hydrocarbons (PAHs) and
26 nitroarenes. These components are of greatest concern because of their demonstrated or
27 suspected activity as procarcinogens or carcinogens in laboratory animals and their universal
28 occurrence in diesel emissions. It is also well known that PAH biological reactivity and
29 carcinogenicity are dependent on their metabolic conversion [reviewed by Conney (1982)].
30 The mechanism of action of many PAH carcinogens has been attributed to the reactivity
31 of certain metabolic intermediates with cellular macromolecules and the subsequent formation
December 1994 10-1 DRAFT-DO NOT QUOTE OR CITE
-------
1 of DNA adducts. The organics adsorbed to diesel exhaust particles may become available
2 for biotransformation to known reactive intermediates, and macromolecular binding of these
3 metabolites has been demonstrated.
4 Except for some of the DNA adduct studies, the available data base does not allow for
5 a definitive discussion of the specific mechanism of carcinogenic action for these compounds
6 relative to diesel exhaust specifically but rather is approached from the standpoint of the
7 chemicals per se. Some of the data are derived primarily from in vitro studies that were not
8 specifically concerned with the potential carcinogenicity of diesel exhaust but are relevant
9 because the compounds of concern are known components of diesel emissions.
10 Several long-term inhalation studies have provided evidence for carcinogenicity and
11 tumorigenicity of whole diesel exhaust in animals (Heinrich et al., 1986; Iwai et al., 1986;
12 Mauderly et al., 1987). Over 100 carcinogenic or potentially carcinogenic components have
13 been specifically identified in diesel emissions, including various PAHs and nitroarenes such
14 as 1-nitropyrene (1-NP) and dinitropyrenes (DNPs). These compounds are adsorbed to the
15 carbon core of the paniculate phase of the exhaust and, upon desorption, may become
16 available for biological processes such as metabolic activation to mutagens. Among
17 compounds identified from diesel exhaust are B[a]P, dibenz[a,/z]anthracene, pyrene,
18 chrysene, and nitroarenes such as 1-NP, 1,3-DNP, 1,6-DNP, and 1,8-DNP, all of which are
19 mutagenic, carcinogenic, or implicated as procarcinogens or cocarcinogens (Stenback et al.,
20 1976; Weinstein and Troll, 1977; Thyssen et al., 1981; Pott and Stober, 1983; Howard
21 et al., 1983; Hirose et al., 1984; Nesnow et al., 1984; El-Bayoumy et al., 1988).
22
23 10.1.1 Metabolism and Disposition of Benzo[a]pyrene
24 It is generally recognized that B[a]P is an activation-dependent carcinogen, with the
25 activated metabolites forming covalent DNA adducts (Boyland, 1980). The reactions
26 responsible for this activation are mediated by the cytochrome P-450 monooxygenases and
27 are known to occur in multiple tissues and in different species. The activation proceeds
28 through Phase I oxidative and hydrolytic reactions, which result in the formation of the
29 ultimate carcinogenic metabolite, B[a]P 7,8-dihydrodiol 9,10-epoxide. Specifically, B[a]P
30 undergoes a mixed function oxidase (MFO)-mediated epoxidation to form B[a]P-7,8-oxide,
31 which, in turn, is subjected to an epoxide hydrolase-mediated hydrolysis resulting in the
December 1994 10-2 DRAFT-DO NOT QUOTE OR CITE
-------
1 stereoisomeric diols, (+)-B[a]P 7,8-dihydrodiol, and (-)-B[a]P 7,8-dihydrodiol. The
2 diasterioisomeric forms of B[a]P 7,8-diol 9,10-epoxide are derived following another
3 P-450-mediated reaction.
4 The principal source of environmental B[a]P is its association with airborne particles
5 such as those generated by diesel engines and coke ovens (U.S. Environmental Protection
6 Agency, 1985). Therefore, understanding the metabolism of chemical carcinogens such as
7 B[fl]P is instrumental in providing a complete understanding of diesel emission-induced
8 carcinogenesis. However, it is beyond the scope of this document to review exhaustively the
9 literature regarding the metabolism of these compounds. The respiratory tract metabolism of
10 particle-associated PAHs has been summarized by Sun et al. (1988a)
11 Relative to diesel exhaust carcinogenicity, several studies have examined the
12 metabolism and disposition of constituents such as E[a]P. Mitchell (1982), subjected
13 24 male F344 rats to nose-only inhalation of 3H-B[a]P aerosol (500 mg/m3) for 60 min.
14 High levels of radiolabel were detected in the trachea, lungs, and turbinates. Based on
15 measurement of the radiolabel, biphasic clearance was noted with half-time (t1/2) values of
16 2 to 3 h and 25 to 56 h. Absorption by the lungs and systemic distribution was demonstrated
17 by the presence of radiolabel in soft tissues, such as the liver, kidney, gastrointestinal tract,
18 spleen, brain, and testes. The majority of the radiolabel in these tissues was removed after
19 2 days, and the major route of excretion was in the feces. The significance of this study is
20 the demonstration of rapid absorption and systemic distribution of B[a]P and potential
21 metabolites following inhalation exposure.
22 Metabolism of intratracheally instilled B[a]P (1.0 /xg) in strain A/J mice exposed to
23 diluted diesel exhaust (8 h/day, 7 days/week for 9 mo) was reported by Tyrer et al. (1981).
24 The radiolabel (14C or 3H) was rapidly distributed throughout the body within 2 h. The
25 highest levels were detected in the lungs, liver, and gastrointestinal tract. Only trace levels
26 were detected in the gastrointestinal tract 168 h after administration. A companion study
27 (Cantrell et al., 1980) examined the effects of prior diesel exhaust exposure on in vivo B[a]P
28 metabolism in the aforementioned mice. Homogenates of lung, liver, and testes were
29 obtained from five mice sacrificed at 2, 24, or 168 h after B[a]P instillation.
30 High-performance liquid chromatography (HPLC) analysis detected free B[a]P and
31 nonconjugated primary metabolites, and sulfate, glucuronide and glutathione conjugates in
December 1994 10-3 DRAFT-DO NOT QUOTE OR CITE
-------
1 each of the tissues. The occurrence of primary and secondary B[a]P metabolites in all three
2 tissues was verified. The major hepatic metabolite was 3-hydroxy-B[a]P. The investigators
3 concluded that the diesel exhaust exposure may qualitatively affect the metabolism of B[a]P
4 but does not significantly affect the distribution of B[a]P.
5 Sun et al. (1984) provided additional information comparing the disposition of particle-
6 adsorbed B[a]P (0.1 wt %) and pure B[a]P following 30-min nose-only inhalation by F344
7 rats. Long-term lung retention (percentage retained after 7 days) of particle-adsorbed
8 3H-B[a]P was approximately 230-fold greater than that for pure 3H-B[a]P. Pulmonary
9 clearance of particle-associated 3H was biphasic, with an initial ty2 of 1 h and a second-phase
10 ti/2 of 18 days, the latter representing clearance of 50% of the initially deposited radiolabel.
11 Clearance of pure B[a]P aerosol was >99% within 2 h and was apparently caused by
12 pulmonary and mucous membrane absorption into the blood rather than by mucociliary
13 clearance and subsequent ingestion (Sun et al., 1982). Of the radiolabel retained in the
14 lungs, 65 to 76% was B[a]P, 13 to 17% was B[a]P-phenol, and 5 to 18% was E[a]P-
15 quinone. Although the Sun et al. (1984) study demonstrated the biotransformation of B[a]P
16 to several metabolites, the epoxide intermediates known to be carcinogenic (Sims et al.,
17 1974; Slaga et al., 1976) were not identified. However, B[a]P-phenol metabolites are
18 reported to be mutagenic (Glatt and Oesch, 1976; Wislocki et al., 1986; Wood et al., 1976).
19 Leung et al. (1988) studied the role of microsomes in the removal and metabolism of
20 B[a]P from diesel exhaust particles. Hepatic and lung microsomal preparations were made
21 from 3-methylcholanthrene-induced F344 rats. 14Carbon-B[a]P was adsorbed to diesel
22 exhaust particles (0.49 /iCi/mg) and incubated with the microsomal preparations. Results
23 indicated that both lung and liver microsomes were capable of removing B[a]P from the
24 exhaust particles and that this capacity was dependent on the lipid content of the microsomes.
25 Only small (<3%) amounts of B[a]P were transferred from the particles, with only 1 to 2%
26 of this being metabolized. Free B[a]P, however, was extensively metabolized by the
27 microsomes to B[a]P-9-10-diol. Relative to the liver microsomes, the lung microsomes
28 exhibited an approximate twofold greater efficiency in the transfer of particle-associated
29 E[a]P.
30 Bond et al. (1984) demonstrated metabolism of particle-associated B[a]P and free B[a]P
31 by alveolar macrophages (AMs). B[a]P-9,10-diol and B[0]P-7,8-diol were identified in the
December 1994 10-4 DRAFT-DO NOT QUOTE OR CITE
-------
1 culture media, and B[a]P-7,8-diol and B[a]P-4,5-diol were detected in the cellular extracts.
2 Additionally, small amounts of B[0]P phenols and B[a]P quinones were detected in both the
3 cells and the media. The total amount of metabolites from both the cells and media were
4 increased with increasing incubation time up to 48 h. However, use of B[a]P in solution or
5 B[a]P coated onto diesel exhaust particles did not alter the total amount of metabolites
6 produced by the macrophages over a 24-h incubation period. Alveolar macrophage-mediated
7 metabolism of particle-associated B[0]P is especially relevant considering that macrophages
8 are instrumental in sequestering and transporting diesel exhaust particulate matter in the
9 lungs. Although this investigation points to the ability of the AMs to metabolize B[a]P
10 associated with diesel particles, Chen and Vostal (1982) have reported that aryl hydrocarbon
11 hydroxylase (AHH) in AMs is decreased after in vivo exposure to diesel exhaust. Whether
12 such diesel-associated decreases in AM enzymatic activity is counterbalanced by increases in
13 the AM population size in response to diesel particle deposition (White and Garg, 1981) is
14 unknown. Although it is known that human AMs contain AHH activity (McLemore et al.,
15 1981), and that they can metabolize E[a]P (Harris, 1985), comparative studies of the AHH
16 activities in rat, hamster, and human AMs could contribute toward determining the
17 relationship such activity may have on the development of lung tumors.
18 Even though the AMs appear to contain the bulk of diesel particles deposited in the
19 lung during chronic exposures, other cell types may also participate in the sequestration
20 and/or metabolic activation of carcinogenic agents. The ability of lung epithelial cells to
21 sequester diesel exhaust particles was reported by White and Garg (1981). Furthermore,
22 significant metabolism of B[a]P by rat Type II alveolar epithelial cells was reported by Bond
23 et al. (1983). In this study, a lung epithelial cell line (LEG) was shown to metabolize B[a]P
24 to B[a]P-7,8-diol and B[a]P-9,10-diol, the latter accounting for 80% of the total B[a]P
25 metabolites. Small quantities of glucuronide conjugates of B[a]P-7,8-diol and
26 9-hydroxy-B[a]P were detected. Preexposure of the cells to diesel exhaust particle extract,
27 benz[a]anthracene, or coal gas condensate increased rates of covalent binding of radioactivity
28 to macromolecules twofold to fivefold. It was also found that pretreatment of LEG with
29 diesel exhaust particle extract produced a threefold increase in covalent binding of 14C-B[fl]P.
30 Compared with the AMs that were examined in the aforementioned study, the rat Type II
31 cells showed approximately a ten times greater ability to metabolize B[a]P.
December 1994 10-5 DRAFT-DO NOT QUOTE OR CITE
-------
1 Under healthy conditions, the Type II cells represent about 12 to 16% of all cells in the
2 pulmonary epithelium of mammalian lungs and account for approximately 4 to 9% of the
3 cells in the lungs (Crapo et al., 1983). Alveolar macrophages, on the other hand, account
4 for approximately 4 to 9% of the cells in the pulmonary region (Crapo et al., 1983).
5 In terms of their relative availability and existing information on their relative abilities to
6 metabolize B[a]P, the Type II cells may play an even more important role in metabolically
7 activating PAH than the AMs, assuming PAH as a substrate is available to them (e.g.,
8 extraction of PAH from diesel particles by AMs and the subsequent release of PAH or
9 metabolically susceptible metabolites of PAH at Type II cell sites). The Type II cell
10 hyperplasia observed after the deposition of diesel and other types of particles (White and
11 Garg, 1981; Lee et al., 1986; Lee et al., 1988; Plopper et al., 1983) seemingly would favor
12 a prominent role for these cells in producing activated PAH metabolites. Another cell type
13 that may be important in the metabolism of PAH to ultimate carcinogens is the nonciliated
14 bronchiolar cell. These cells are relatively rich in chemical metabolizing enzymes and, being
15 also in a region of the respiratory tract where clearance of material would be relatively fast,
16 may receive exposure via mucus to organics that have desorbed in the pulmonary region.
17 The respiratory tract cytochrome P-450 system, for example, is present in Type II cells but it
18 is not as concentrated in this epithelial cell type as it is in the nonciliated bronchiolar cell
19 (Boyd, 1984). It is worthy to note that bronchoalveolar adenomas that develop following
20 diesel exposure have been found to resemble both Type II and nonciliated bronchiolar cells
21 (Mauderly et al., 1987). Like the Type II cells, the nonciliated bronchiolar cells are viewed
22 as not being important in the phagocytosis of particles that deposit in the lung.
23 As previously indicated, any metabolism of procarcinogens by these cells probably involves
24 the preextraction of carcinogen(s) in the extracellular lining fluid and/or in other endocytic
25 cells.
26 It is evident from the preceding studies that B[a]P adsorbed to diesel exhaust particles
27 that are deposited in the respiratory tract can be readily distributed throughout much of the
28 organism via absorption from the lung and transport by the mucociliary escalator to the
29 gastrointestinal tract. The current data base appears to support the contention that particle-
30 associated B[a]P can ultimately be metabolized by AMs and/or Type II cells to reactive
31 intermediates following deep lung deposition of these particles.
December 1994 10-6 DRAFT-DO NOT QUOTE OR CITE
-------
1 10.1.2 Carcinogenic Mechanism of Benzo[a]pyrene
2 As a result of PAH metabolism studies such as those conducted on B[a]P, theories have
3 been proposed regarding the molecular mechanism by which activated intermediates express
4 their genotoxic effects. Benzo[a]pyrene served as the model for the "bay-region" concept
5 summarized by Jerina et al. (1980). This proposed mechanism would also be applicable to
6 such compounds as benz[a,h]anthracene, which is a potent carcinogen also known to occur as
7 a diesel exhaust combustion product.
8 Briefly, this concept states that compounds derived from an angular benz[a]anthracene
9 nucleus may undergo epoxidation, and if the resulting epoxides are located in the bay region,
10 they will be better alkylating agents and, therefore, have a greater genotoxic potential. The
11 chemical reactivity of these bay-region epoxides is positively correlated with biological
12 reactivity of these compounds.
13 Based on the assumption that DNA adduct formation is a critical step in the initiation of
14 carcinogenesis (Harris, 1985), increased residence time of PAHs in the lung would increase
15 the opportunity for metabolism and subsequent adduct formation. This would be especially
16 important if association of the PAHs with the soot particles and their slow release from these
17 particles contributed to this increased residence time. Therefore, comparison of adduct
18 formation by B[a]P alone to that of particle-associated B[a]P is important for understanding
19 possible mechanisms of diesel exhaust carcinogenicity.
20 An experiment was undertaken to test the hypothesis that inhalation of B[a]P associated
21 with carbon black (CB) particles would increase the levels of DNA adducts compared with
22 inhalation of pure B[a]P (Wolff et al., 1989). The DNA modification was measured using
23 the 32P-postlabeling method recently developed by Randerath et al. (1985). The high
24 sensitivity («1 adduct in 1010 bases) of this technique (Reddy and Randerath, 1986) made
25 possible measurement of the low levels of DNA adducts resulting from repeated inhalation
26 exposures to 14C-B[a]P aerosols (2 mg/m3), 14C-B[a]P (2 mg/m3) adsorbed to CB particles
27 (97 mg/m3) (B[a]P/CB), or filtered air. Total 14C levels in the lung (a nonspecific indicator
28 of reactive and nonreactive B[a]P metabolites, free B[a]P, and particle bound Bfrz]P) were
29 100-fold greater following exposure to B[a]P/CB than following exposure to B[a]P alone.
30 The levels of total DNA adducts or the B[a]P diol-epoxide(BPDE)-DNA adduct in the
31 lung were not significantly different whether the rats were exposed to pure B[a]P or
December 1994 10_7 DRAFT-DO NOT QUOTE OR CITE
-------
1 B[a]P/CB. However, association of B[a]P with CB resulted in the formation of unidentified
2 lung adducts that were not seen in DNA from lungs of rats exposed to pure B[a]P. It is
3 possible that the adducts seen only in the B[a]P/CB exposures may play a role in the
4 potential tumorigenic effect of particle-associated B[a]P. Reasons for the discrepancy
5 between particle effects on total DNA adducts and retention of 14C include the possibility that
6 the kinetics for formation and decline of DNA adducts are different from those of total bound
7 14C. As a consequence, long-term retention of total B[a]P and metabolites in the lung may
8 not be a good marker for adduct formation.
9 There were clear differences in the kinetics of the buildup and decline of DNA adduct
10 levels and total 14C for rats exposed to B[a]P/CB. The t1/2 for the decline of total 14C was
11 approximately tenfold faster than that for the decline in levels of DNA adducts for rats
12 exposed to B[<2]P/CB. Previous work has shown that at 1 day or later after the end of single
13 exposures to B[a]P or B[a]P/CB, most of the 14C present was bound to total macromolecules
14 (Sun et al., 1988b), presumably largely, non-DNA protein. Thus, this information in
15 combination with the current data suggests that decline or repair of DNA adducts is
16 considerably faster than that of protein turnover. Following repeated exposures, this would
17 be expected to lead to increased buildup of 14C in the lung relative to DNA adducts. The
18 tj/2 values for decline in DNA adducts observed in the current work are similar to the
19 t1/2 values of approximately 4 weeks reported for B[a]P metabolite-DNA adducts in the lungs
20 of A/HeJ and C57B1/6J mice (Stowers and Anderson, 1985). Protein turnover is generally
21 on longer time scales than the aforementioned t1/2 values.
22 It appears that long-term retention of 14C radiolabel in the lung may not be as important
23 as previously suspected, at least with respect to indicating DNA damage. The 14C binding
24 levels and DNA adducts were not closely related, and it is clear from these results that DNA
25 adduct levels cannot be predicted from total 14C levels. This observation is consistent with
26 the work of Morse and Carlson (1985) who observed that binding levels of 3H with lung
27 protein were greater than levels of 3H to lung DNA 6 h after administration of oral H-B[a]P
28 to mice. They also found that 3H binding to protein was more persistent than 3H binding to
29 DNA.
30 Caution should be used in interpreting the results from short-term exposures in regard
31 to possible implications for long-term exposures when carcinogenicity might be observed.
December 1994 10-8 DRAFT-DO NOT QUOTE OR CITE
-------
1 The same pattern of results seen after 12 weeks might not continue after many months of
2 exposure. The adduct levels were higher in the rats exposed to B[a]P/CB than B[a]P after
3 12 weeks of exposure, and so it is possible that this difference might become greater with
4 continued exposure. In addition, the different adduct patterns between the B[a]P/CB and
5 B[a]P exposures may indicate that other adducts besides the BPDE-DNA adduct are
6 important in potential carcinogenic effects of B[a]P/CB exposures. Another factor to
7 consider is the possible influence of a chronic inflammatory response, cell injury, and cell
8 proliferation, which accompany long-term exposures to inhaled insoluble particles (Morrow,
9 1986). Such responses are generally greater after prolonged exposure than those in the
10 current 12-week exposure. These responses might be factors in progression to tumors in
11 long-term inhalation exposures of rodents, when large lung burdens of particles accumulate
12 (Morrow, 1986), and in the increased incidence in tumors, when B[a]P is merely mixed with
13 Fe2O3 particles versus adsorbed onto the particle (Saffiotti et al., 1965).
14
15 10.1.3 Metabolism and Disposition of Nitroarenes
16 Diesel engine emissions contain a large number of components including an extensive
17 list of nitroarenes. Quantitatively, the nitroarenes represent a relatively small contribution to
18 the overall PAH component of diesel engine emissions. However, with respect to the
19 carcinogenic potential of diesel exhaust, some of the nitroarenes (e.g., 1-NP, 4-NP,
20 6-nitrochrysene, and some DNPs) are of concern because of their known or suspected
21 carcinogenic activity and their high mutagenic activity in some test systems (Manabe et al.,
22 1985; International Agency for Research on Cancer, 1989). Within the scope of this
23 document, it is inappropriate to review of all of the studies regarding the carcinogenicity,
24 metabolism, and mechanism of action of these various nitroarenes. Therefore, emphasis has
25 been placed on those nitroarenes considered by the International Agency for Research on
26 Cancer (1989).
27 1-Nitropyrene, a genotoxic and carcinogenic nitro-substituted organic, is a particle-
28 associated component of diesel exhaust (Pitts et al., 1982; Schuetzle et al., 1982; King,
29 1988). As with B[0]P, several investigators have studied the metabolism and disposition of
30 1-NP both in free form and in association with diesel exhaust particles.
December 1994 10_9 DRAFT-DO NOT QUOTE OR CITE
-------
1 Bond and Mauderly (1984) made quantitative measurements of 1-NP metabolism and
2 macromolecular covalent binding in the isolated-perfused rat lung. The study verified
3 oxidation, reduction, acetylation, and conjugation biotransformation of 1-NP by the lung,
4 with oxidation being the major process. The major metabolites were 3-, 6-, and
5 8-hydroxynitropyrene. The overall metabolism of 1-NP was increased by prior exposure of
6 the rats to the mixed-function oxygenase (MFO) inducer 3-methylcholanthrene (3-MC) but
7 not to phenobarbital. This 3-MC-induced increase hi 1-NP metabolism and a parallel
8 increase in macromolecular covalent binding suggests that this pathway may be responsible
9 for the observed covalent binding.
10 It is also noteworthy that hydroxynitropyrenes have been shown to be mutagenic using
11 the Ames assay (El-Bayoumy and Hecht, 1983).
12 Exposure of rats to diesel exhaust (7.4 mg/m3) for 7 h/day, 5 days/week for 4 weeks
13 resulted in twofold increases in the rates of nitropyrene metabolism in nasal tissue and in
14 isolated perfused lungs from these animals (Bond et al., 1986). High-performance liquid
15 chromatography analysis of ethyl acetate-extractable 1-[14C]NP metabolites indicated that the
16 major metabolites were 3-, 6-, and 8-hydroxy-l-aminopyrene and 4,5-dihydro-4,5-dihydroxy-
17 1-nitropyrene. Furthermore, a fourfold increase in 14C covalently bound in the lungs of
18 these rats was detected. The increase in 1-NP metabolism was not observed for rats of
19 lower-exposure (0.35 or 3.3 mg/m3) groups or clean air controls. The data from this study
20 indicate that exposure to diesel exhaust paniculate matter at concentrations of 7.4 mg/m3
21 significantly alters the metabolism and subsequent covalent binding of nitropyrene.
22 Bond et al. (1986) also examined the metabolism and deposition of free and particle-
23 associated 1-NP in F344 rats. Results of the work indicated that the urinary and fecal
24 excretion of 14C-1-NP was not altered by exposure to the pure form or to that adsorbed on
25 diesel exhaust particles. Pure 1-NP was more efficiently absorbed in the lung than was 1-NP
26 coated onto diesel exhaust particles, and, therefore, greater lung retention was noted for
27 particle-adsorbed 1-NP. However, no significant difference between the two forms of 1-NP
28 was noted for extrapulmonary tissue distribution or metabolic profiles. Analysis of excreta
29 and tissues indicated that 1-NP is rapidly metabolized by the lungs or metabolized by other
30 tissues following translocation from the lungs. For both 1-NP forms, small amounts of
31 6- and 8- hydroxyacetylaminopyrene were detected in the lungs, suggesting pulmonary
December 1994 10-10 DRAFT-DO NOT QUOTE OR CITE
-------
1 oxidation, reduction, and conjugation of the parent compound. The demonstration of
2 pulmonary metabolism of 1-NP and greater retention of 1-NP when adsorbed to diesel
3 exhaust particles may be significant relative to the dose to the lungs of both parent compound
4 and metabolites.
5 Ball and King (1985) administered [14C]1-NP to rats intraperitoneally, orally, or by
6 intratracheal instillation of vapor-phase-coated diesel exhaust particles (380 fig [14C]l-NP/g;
7 5 mg/rat). Over 50% of the radiolabel was recovered (20 to 30% in the urine and 40 to
8 60% in the feces) within 24 h, regardless of the route of administration. The metabolic
9 profile and elimination kinetics were similar for all routes of administration. The principle
10 urinary metabolite (representing 15 to 25% of the total urinary 14C) was 6-hydroxy-A^-acetyl-
11 1-aminopyrene (6-OH-NAAP), a compound with demonstrated S-9 dependent mutagenic
12 activity in Salmonella strain TA98. Gut flora was shown to be necessary for the formation
13 of 6-OH-NAAP, for the observed enterohepatic circulation of metabolites excreted in the
14 bile, and for excretion of mutagenic activity in the urine. That intestinal microorganisms
15 may alter the metabolites of 1-NP and facilitate their reabsorption was also reported by
16 Medinsky et al. (1985). Accumulation of 14C and diesel exhaust particles was detected in the
17 lungs and gastrointestinal tract 24 h after intratracheal administration, thereby attesting to the
18 importance of mucociliary transport and distribution of particles and their adsorbed
19 components. Based on these results and previous in vitro studies (King et al., 1983)
20 demonstrating 1-NP binding to macromolecules, the authors note the possible risk to the
21 gastrointestinal tract and lungs relative to 1-NP.
22 The biotransformation of 1-NP by intestinal microflora of humans, rats, and mice was
23 also reported by King et al. (1990). Metabolites including 1-aminopyrene, W-acetyl-1-
24 aminopyrene, N-formyl- 1-aminopyrene, and two unkown compounds were detected. Except
25 for N-formy 1-1 -aminopyrene, all of the metabolites as well as the residual parent compound
26 were mutagenic in reverse-mutation assays.
27 Medinsky et al. (1988) examined the distribution of covalently bound 14C following the
28 administration of [14C]-1-NP to rats by nose-only inhalation (340 ng/L for 1 h) or gavage
29 (4.2 /ig in 0.5 mL saline). Ninety-six hours after administration, covalently bound 14C was
30 highest in the kidney, followed by the liver and lungs, respectively. These findings suggest
December 1994 10_n DRAFT-DO NOT QUOTE OR CITE
-------
1 that the kidney may be a potential target organ for 1-NP, whether the route of administration
2 is inhalation or gavage.
3 Howard et al. (1986) studied the binding of intratracheally instilled nitropyrenes and
4 B[a]P to mouse lung DNA following preexposure to intratracheally instilled doses of the
5 putative inducing agents, B[a]P, dichloromethane extract of diesel exhaust, or 1-NP. The
6 results indicated that 1-NP was a potent DNA-binding agent even in the absence of enzyme
7 induction and that this potency was increased following B[a]P exposure. Dinitropyrene
8 (a mixture of the 1,3-, 1,6-, and 1,8- isomers) was also a potent lung DNA-binding agent,
9 with and without the inducers. Benzo[a]pyrene was not as potent a binding agent.
10 Preexposure to the diesel exhaust extract, but not to B[a]P, resulted in increased DNA
11 binding of B[0]P. Pretreatment with the dichloromethane extract of diesel exhaust failed to
12 increase the DNA binding of the nitropyrenes. The significance of this report is the
13 demonstration that exposure to diesel exhaust may potentiate the DNA binding of some of its
14 components.
15 King (1988) provided information relating the metabolism of nitropyrenes and their
16 carcinogenic potential. Briefly, DNPs were found to be much more carcinogenic than 1-NP.
17 The cytosolic enzymes of the rat mammary gland activated DNPs by monoreduction to
18 hydroxylamine followed by O-acetylation. Reduction and acetylation pathways for DNPs and
19 subsequent DNA adduct formation were detected in intact cells. However, the intact
20 mammary cells metabolized 1-NP primarily through oxidative pathways.
21 Physiologically based pharmacokinetic modeling was used by Medinsky and co-workers
22 to simulate the disposition of 1-NP after its inhalation or ingestion by rats (Medinsky et al.,
23 1989). The model utilized physiologically realistic organ volumes, organ blood flow values,
24 and tissue/blood partition coefficients for 1-NP and its metabolites. 1-Nitropyrene displayed
25 a higher affinity for blood than for lung tissue (lung/blood partition coefficient of 0.5),
26 whereas the liver/blood and kidney/blood partition coefficients were equal to 1, implying that
27 perfusion rate alone could account for removal of 1-NP from these organs. The partition
28 coefficients for the 1-NP metabolites were less than that of 1-NP, indicating that these would
29 more efficiently partition into the blood and be excreted via the urine or bile. These findings
30 are consistent with the reported rapid excretion of 1-NP metabolites.
December 1994 10-12 DRAFT-DO NOT QUOTE OR CITE
-------
1 The International Agency for Research on Cancer (1989) classifies 4-nitropyrene (4-NP)
2 as a possible human carcinogen and identifies it as a component of diesel exhaust. However,
3 little information is available regarding the metabolism of this compound. One report (Fu et
4 al., 1986) reviewed in International Agency for Research on Cancer (1989) reported the
5 in vitro biotransformation of 4-NP to 4-nitropyrene 9,10-dione, 8-hydroxy-4-nitropyrene-l,6-
6 hydroquinone by rat liver microsomes. The report also indicated that this metabolite was
7 mutagenic.
8 6-Nitrochrysene has been shown to be metabolized to 6-nitrosochrysene,
9 6-aminochrysene, Af-formyl-6-aminochrysene, and 6-acetylaminochrysene by anaerobic
10 bacteria isolated from human feces, which implies such metabolites may be formed in the
11 lower gastrointestinal tract (El-Bayoumy and Hecht, 1984). Manning et al. (1988) reported
12 the reduction of 6-nitrochrysene to 6-aminochrysene by cultures of intestinal bacteria from
13 rats and mice and pure cultures of anaerobic bacteria. The in vivo metabolism of
14 6-nitrochrysene hi preweanling mice was implied by the detection of a DNA adduct in the
15 lung and liver that corresponded to a DNA adduct found in microsomal incubations
16 containing calf thymus DNA and 6-aminochrysene trans dihydrodiol (Delclos et al., 1988).
17 The in vitro biotransformation of 1,3-DNP was verified by Djuric et al. (1985),
18 wherein the compound was reduced to l-amino-3-nitropyrene, l-nitro-3-nitrosopyrene, and
19 1,3-diaminopyrene by rat and dog liver cytosol in an argon atmosphere. The addition of
20 acetyl coenzyme A (Co-A) to the liver cytosol incubations resulted in the formation of an
21 additional metabolite, l-acetylamino-3-nitropyrene. Although intact rat mammary gland cells
22 did not metabolize 1,3-dinitropyrene, in vitro incubations using cytosolic preparations from
23 this type of cell resulted in the formation of l-amino-3-nitropyrene and 1,3-diaminopyrene
24 (King et al., 1986; Imaida et al., 1988).
25 Several studies have reported on the in vivo and in vitro metabolism of
26 1,6-dinitropyrene. Single ip injections of [3H]1,6-DNP (33.6 pig) in preweanling CD mice
27 resulted in the detection of an N-(deoxyguanosin-8-yl)-l-amino-6-nitropyrene adduct in lung
28 and liver tissue (Delclos et al., 1987b). Following ip administration of [3H]1,6-DNP
29 (200 /ig/kg) to male Sprague-Dawley rats, this same adduct was detected in the urinary
30 bladder, liver, and kidney, and in mammary epithelium (Djuric et al., 1988). In vitro
31 studies have also demonstrated the biotransformation of 1,6-DNP to l-amino-6-nitropyrene,
December 1994 10-13 DRAFT-DO NOT QUOTE OR CITE
-------
1 l-nitro-6-nitrosopyrene and 1,6-diaminopyrene by rat and dog liver cytosol (Djuric et al.,
2 1985) and also by rat mammary gland cytosol (King et al., 1986).
3 The International Agency for Research on Cancer (1989) considers 1,8-DNP to be a
4 possible human carcinogen. In vivo and in vitro studies have demonstrated the formation of
5 several metabolites. Following oral administration of 0.3 mg of 1,8-DNP to male CD rats,
6 N,N'-diacetyl-l,8-diaminopyrene, 1,8-diaminopyrene, l-acetylamino-8-nitropyrene, and
7 l-amino-8-nitropyrene were detected in the feces (Heflich et al., 1986). Additional,
8 unidentified polar metabolites were also detected. The fact that only l-amino-8-nitropyrene
9 and the polar metabolites were detected for germ-free rats treated similarly suggests the
10 involvement of intestinal microflora in the in vivo biotransformation of 1,8-DNP. In this
11 same study, incubation of 1,8-DNP with rat or dog liver cytosol preparations in an argon
12 atmosphere produced l-amino-8-nitropyrene, l-nitro-8-nitrosopyrene, and 1,8-diaminopyrene.
13 Incubations of human liver cytosol and 1,8-DNP resulted in the production of l-amino-8-
14 nitropyrene, and rat mammary gland cytosol incubations under anaerobic conditions produced
15 l-nitro-8-nitrosopyrene, l-amino-8-nitrosopyrene, and 1,8-diaminopyrene (King et al., 1986;
16 Imaidaetal., 1988).
17 Moller et al. (1988) have reviewed the in vivo metabolism of 2-nitrofluorene, a
18 constituent of diesel exhaust that the International Agency for Research on Cancer (1989)
19 classifies as possibly carcinogenic to humans. The metabolites 2-aminofluorene,
20 2-acetylaminofluorene, and 2-formylaminofluorene were detected in the urine of rabbits and
21 in the feces of rats following oral administration of 2-nitrofluorene (Tatsumi and Amano,
22 1987). Studies using [14C]2-nitrofluorene administered orally to rats demonstrated the
23 formation of N-, 1-, 3-, 5-, 7-, 8-, and 9-hydroxy-2-acerylaminofluorene. The International
24 Agency for Research on Cancer (1989) noted that N-hydroxy-2-acetylaminofluorine and
25 9-hydroxy-2-acetylaminofluorene were known to be carcinogenic in animals. Moller et al.
26 (1987) reported that the isolated perfused rat lung metabolizes 2-nitrofluorene to
27 9-hydroxy-2-nitrofluorene and an unidentified hydroxylated nitrofluorene.
28 The preceding studies have shown that some of the nitroarenes known to be constituents
29 of diesel exhaust may undergo biotransformation to various metabolites, some of which are
30 known to be carcinogenic to animal species. Such data may become more relevant as a
December 1994 10-14 DRAFT-DO NOT QUOTE OR CITE
-------
1 complete understanding is obtained regarding the desorption of these compounds from the
2 soot particle and their subsequent availability for biotransformation processes.
3
4 10.1.4 Carcinogenic Mechanism of Nitroarenes
5 Although the nitroarenes quantitatively represent a relatively small portion of the PAH
6 component of diesel engine exhaust, their contribution to the potential carcinogenicity of
7 diesel engine emissions deserves consideration. In the previous section, information was
8 presented regarding the in vivo and in vitro metabolism of various nitroarenes considered to
9 be possible human carcinogens. The fact that some of these metabolites have been shown to
10 form DNA adducts in animal studies and are mutagenic in several test systems warrants their
11 inclusion in assessing possible mechanisms of diesel-exhaust-induced carcinogenicity.
12 In a study by Sato et al. (1986), sc injection (4.0 mg total dose) of 1,3-NP, 1,6-NP, or
13 1,8-NP into male F344 rats resulted in a 100% (10 of 10) sarcoma incidence at the injection
14 site. After 320 days of observation, the dimethyl sulfoxide controls exhibited a tumor
15 incidence of 0% (0 of 20). No tumors were found in rats receiving 4 or 40 mg of 1-NP.
16 Transforming activity was observed for the DNA from 4 of 11 DNP-induced tumors,
17 suggesting oncogene activation.
18 The absence of tumor formation by 1-NP and the positive results for the DNPs from
19 the previous study are consistent with the findings of King (1988). In this study, female CD
20 and F344 rats were administered 1,3-, 1,6-, or 1,8-DNP, or 1-NP subcutaneously,
21 intraperitoneally, or intragastrically. Based on incidences of malignant histiocytomas,
22 mammary gland tumors and leukemias, the DNPs were much more carcinogenic than 1-NP.
23 Specifically, the potency order was 1,6-DNP > 1,8-DNP > 1,3-DNP > 1-NP. Oral
24 intubation was relatively ineffective in tumor induction by these compounds. Cytosolic
25 enzymes of the rat mammary gland were capable of activating the DNPs to reactive
26 intermediates that formed tRNA adducts. The adduct formation was of the same relative
27 order as the carcinogenic potential and was catalyzed by rat mammary cytosol. In addition
28 to demonstrating the potential of DNPs to induce three tumor types in rats, this study
29 provided data affirming the importance of metabolic activation of the DNPs, the quantitative
30 variability among these isomers relative to their carcinogenic activity, and the susceptibility
31 of a tissue distant from the site of administration to the carcinogenic potential of these
December 1994 10-15 DRAFT-DO NOT QUOTE OR CITE
-------
1 compounds following biotransformation. The data support the conclusion that the DNPs
2 (especially 1,6- and 1,8-NP) are reduced by mammalian enzymes to reactive hydroxylamines,
3 which are in turn activated by acetyl Co-A to potential carcinogens.
4 Maher et al. (1988) examined the metabolism of 1-NP and DNPs in cultured human
5 fibroblasts. The results of their experiments using these diploid cells indicated that 1-NP
6 underwent bioactivation to a form that produced stable covalent DNA adducts. The
7 mutagenic effect of 1-NP was correlated with the cytotoxic effect, which in turn correlated
8 directly with the number of DNA adducts. It was also reported that 1-nitrosoaminopyrene
9 (1-NOP), a major metabolite of 1-NP was similar to 1-NP in its mutagenic response and
10 extent of DNA adduct formation. To reduce survival of the normal (i.e., not repair-
11 deficient) cells to 37%, 25 1-NP or 1-NOP adducts per 106 DNA nucleotides were required.
12 However, comparisons with repair-deficient cells did indicate that nucleotide excision repair
13 protected against the mutagenic and cytotoxic effects of both 1-NP and 1-NOP. The
14 mutagenic potency of both 1-NP and 1-NOP was intermediate between that of N-acetoxy-2-
15 acetylaminofluorene and B[a]P-7,8-diol-9,10-epoxide. Based on Salmonella typhimurium
16 assays, Heflich et al. (1986) reported that the mutagenicity of 1-NOP was 20-fold greater
17 than that of 1-NP.
18 El-Bayoumy et al. (1988) investigated the comparative tumorigenicity of 1-NP and its
19 reduced derivatives, 1-nitrosopyrene and 1-aminopyrene. Results of their tests using
20 Sprague-Dawley rats indicated that 1-NP was considerably more carcinogenic than its
21 reduced derivatives relative to production of mammary adenocarcinomas folio whig
22 administration of these compounds by gavage. These results appear to be hi conflict with
23 those of Wislocki et al. (1986) in which 1-nitrosopyrene produced a greater incidence of
24 hepatic tumors in newborn mice.
25 Roy et al. (1989) suggested that major DNA adducts resulting from the metabolism of
26 1-NP were not the result of simple nitroreduction. In this study, 32P-postlabeling was used
27 to detect multiple adducts in mammary fat pads and livers of rats that had been administered
28 1-NP by gavage. However, DNA adducts resulting from hi vitro nitroreduction
29 cochromatographed with W-(deoxyguanosin-8-yl)- 1-aminopyrene but not with the major
30 adducts found in the mammary fat pads and livers of the treated rats. The investigators
December 1994 10-16 DRAFT-DO NOT QUOTE OR CITE
-------
1 suggested that alternate metabolic pathways, such as ring oxidation or ring oxidation
2 followed by nitroreduction, might be responsible for the formation of these adducts.
3 As described in Section 10.1.3., rat and dog liver cytosol preparations catalyzed the
4 reduction of 1,3-DNP to several products that were shown to bind to exogenous DNA
5 (Djuric et al., 1985). Furthermore, the addition of acetyl Co-A to the incubations resulted in
6 a 19-fold increase in the binding of metabolites to DNA. The involvement of acetyl Co-A
7 was also demonstrated by increased binding of the 1,3-DNP metabolites, l-amino-3-
8 nitropyrene, and 1,3-diaminopyrene to tRNA following its addition to rat mammary gland
9 cytosol incubations (King et al., 1986; Imaida et al., 1988).
10 Djuric et al. (1988) also provided data affirming the correlation between DNA binding
11 of 1-NP or 1,6-DNP and the relative tumorigenicity of the two compounds. Following ip
12 injections of the radiolabeled test compounds, covalent DNA binding was not detected for
13 1-NP treated animals, but the DNA adduct, AKdeoxyguanosin-8-yl)-l-amino-6-nitropyrene
14 was detected in the livers, kidneys, urinary bladder, and mammary glands of rats given
15 1,6-DNP. Induction of nitroreductases failed to increase the DNA binding by 1,6-DNP,
16 suggesting that additional factors such as O-acetylation may affect the observed DNA
17 binding.
18 An analysis of DNA binding of 1-NP metabolites following intratracheal instillation in
19 mice was reported by Mitchell (1988). Several adducts were found in the lung, liver, and
20 kidney at 1 day after administration. Upon HPLC analysis, one adduct, accounting for 20%
21 of the total eluted radioactivity in the lung, coeluted with N-(deoxyguanosin-8-yl)-l-
22 aminopyrene (C8-dG-AP). This adduct was also identified in the liver and kidneys. It was
23 also noted that the C8-dG-aminopyrene adduct remained in the lung for as long as 28 days
24 after administration of the 1-NP, suggesting that the persistence of this adduct and/or others
25 may potentially be associated with the induction of lung tumors. These findings suggest that
26 this adduct may result from the nitroreduction of 1-NP in the lung, liver, and/or kidney.
27 These are especially relevant observations when one considers that the lung is a major route
28 of entry for many materials, including diesel exhaust, that contain nitropyrenes.
29 Based on an increased incidence of mammary tumors in newborn rats, the International
30 Agency for Research on Cancer (1989) considers 4-nitropyrene to be possibly carcinogenic to
31 humans. The genotoxic effects of 4-nitropyrene are summarized in International Agency for
December 1994 10-17 DRAFT-DO NOT QUOTE OR CITE
-------
1 Research on Cancer (1989), but no data were available regarding DNA adduct formation
2 in vivo. However, because of its structural similarity to 1-nitropyrene, it is likely that
3 4-nitropyrene could be metabolized to intermediates capable of forming DNA adducts.
4 Although the concentration of DNPs in diesel exhaust is low, some DNPs such as
5 1,3-DNP, 1,6-DNP, and 1,8-DNP have been shown to be carcinogenic in animals.
6 Furthermore, DNA adduct formation by metabolites of these compounds has been verified.
7 A more complete review of DNP carcinogenicity appears in International Agency for
8 Research on Cancer (1989).
9 Inhalation studies of 1,3-DNP are limited, but a number of studies have examined its
10 effects following oral and parenteral administration. Subcutaneous administration of
11 1,3-DNP (>99% purity) to 43 newborn female CD rats (weekly injections over an 8-week
12 period for a total dose of 1.0 mg) resulted in a significant increase (5 of 43 rats, p < 0.05)
13 in the incidence of malignant fibrous histiocytomas at the injection site (King, 1988).
14 No such tumors were identified hi the dimethylsulfoxide (DMSO) vehicle control group.
15 A similar study reported by Otofuji et al. (1987) used 6-week-old male BALB/c mice to
16 which 1,3-DNP (>99% purity) was administered subcutaneously at dose of 0.05 mg
17 (in DMSO) weekly for 20 weeks. No sc tumors were detected in the treated animals, but the
18 observation period lasted for a maximum of only 60 weeks. Lung, liver, and spleen tumors
19 were identified in the 1,3-DNP-treated mice, and the incidence was significantly greater than
20 that for DMSO controls. In vitro studies have verified DNA binding by 1,3-DNP
21 metabolites, the extent of which could be increased 19-fold if acetyl Co-A was added to the
22 incubation preparations (Djuric et al., 1985).
23 Both in vivo and in vitro experiments have confirmed DNA adduct formation by
24 metabolites of 1,6-DNP. Twenty-four hours after a single ip dose of [3H] 1,6-DNP to
25 preweanling mice, Delclos et al. (1987b) detected the adduct ^V-(deoxyguanosin-8-yl)-l-
26 amino-6-nitropyrene in lung and liver DNA. Male Sprague-Dawley rats receiving an ip dose
27 of [3H] 1,6-DNP exhibited DNA adducts hi the mammary epithelium, liver, lung, kidney, and
28 urinary bladder. Consistent with these data, Wislocki et al. (1986) reported the development
29 of liver tumors (adenomas and carcinomas) in mice treated with 1,6-DNP. King (1988)
30 reported a significantly increased incidence of malignant fibrous histiocytomas hi the
31 peritoneal cavity of rats administered 1,6-DNP intraperitoneally.
December 1994 10-18 DRAFT-DO NOT QUOTE OR CITE
-------
1 Using cytosol preparations from rat or dog liver, Djuric et al. (1985) showed that the
2 resulting 1,6-DNP metabolites (l-amino-6-nitropyrene, l-nitro-6-nitrosopyrene and
3 1,6-diaminopyrene) would bind to exogenous DNA. The addition of acetyl Co-A to the
4 incubations produced an additional metabolite, l-acetylamino-6-nitropyrene, and also greatly
5 increased the extent of its binding to exogenous DNA. The aforementioned metabolites were
6 also detected in rat mammary gland cytosol anaerobic incubations, and their binding to tRNA
7 increased when acetyl Co-A was added to the incubations (King et al., 1986; Imaida et al.,
8 1988).
9 Results of studies using 1,8-DNP were similar to those for 1,6-DNP. The
10 N-(deoxyguanosin-8-yl)-l-amuio-8-nitropyrene adduct was detected in liver and mammary
11 gland cytosol incubations from normal and germ-free CD rats receiving 1.0 fimol of
12 1,8-DNP orally (Heflich et al., 1986). The binding was lower in the germ-free rats
13 suggesting the involvement of intestinal microflora in the production of the adduct forming
14 metabolites. Similar to other DNPs, acetyl Co-A increased the extent of metabolite binding
15 to exogenous DNA by rat and dog liver cytosol incubated under anaerobic conditions (Djuric
16 et al., 1985). As with 1,6-DNP, incubation of rat mammary gland cytosol in the presence of
17 acetyl Co-A resulted in metabolites binding to exogenous tRNA (King et al., 1986; Imaida
18 et al., 1988). Several other test systems, including mouse embryo fibroblasts and Chinese
19 hamster ovary cells, also showed DNA adduct formation by 1,8-DNP metabolites
20 (International Agency for Research on Cancer, 1989).
21 As described in the previous section, 2-nitrofluorene may be metabolized in vivo to
22 products known to be carcinogenic in animal species. However, data are lacking regarding
23 adduct formation by 2-nitrofluorene or its metabolites.
24 Lung and liver adduct formation in preweanling mice parenterally administered
25 6-nitrochrysene was reported by Delclos et al. (1987b). The adducts corresponded to the
26 6-aminochrysene trans-1,2-dihydrodiol adduct produced by microsomal incubations with calf
27 thymus DNA. Incubations using primary cultures of rat liver hepatocytes incubated with
28 6-nitrochrysene formed two DNA adducts, N-(deoxyguanosin-8-yl)-6-nitrochrysene and
29 N-(deoxyguanosin-8-yl)-6-aminochrysene (Delclos et al., 1987a).
30 In summary, both in vivo and in vitro studies have demonstrated the formation of
31 adducts involving 1-NP and its metabolites and metabolites of other nitroarenes known to be
December 1994 10_i9 DRAFT-DO NOT QUOTE OR CITE
-------
1 constituents of diesel exhaust. Although these data do not provide a definitive description of
2 the mechanism of carcinogenic action, the formation of DNA adducts by known constituents
3 of diesel exhaust affirms that interaction with key cellular components may be possible if
4 these components undergo desorption from the soot particles and become available for
5 biotransformation. The ingestion of particle-associated organics via mucociliary transport
6 following inhalation exposure is demonstrated by their metabolic transformation by intestinal
7 microflora and extrapulmonary tissues and subsequent DNA adduct formation.
8
9
10 10.2 PARTICLE EFFECT IN DIESEL EXHAUST-INDUCED
11 CARCINOGENICITY
12 The role of the carbonaceous core (soot particle) and a particle overload effect in the
13 pulmonary carcinogenesis of diesel exhaust is also of concern. Several studies (Vostal, 1986;
14 Kawabata, 1986; Heinrich, 1990; Wolff et al., 1990; Oberdorster and Yu, 1990) have
15 provided data indicating that the carbonaceous core may have a promotional effect related to
16 the ability of the particle to induce chronic inflammation and promote epithelial cell
17 proliferation. More recent work (Nikula et al., 1991, 1994) has shown that carbon black
18 was also carcinogenic in rats exposed to particle concentrations of 2.5 or 6.5 mg/m3 for
19 24 mo. The ramifications of particle overload were discussed more fully in the dosimetry
20 chapter (Chapter 4).
21 A study by Wolff et al. (1990) addressed this topic by comparing the inflammatory
22 responses in rats exposed to diesel exhaust (10 mg/m3) or CB particles (10 mg/m3).
23 Although the level of lung DNA adducts was slightly higher for diesel exhaust exposure,
24 both exposures resulted in inflammatory responses, as determined by increased numbers of
25 neutrophils and macrophages and increased acid proteinase in the bronchoalveolar lavage
26 fluid.
27 Oberdorster and Yu (1990) evaluated the significance of a particle effect in the
28 tumorigenic response of the lung to diesel exhaust exposure. Using data from studies
29 examining the effects of long-term inhalation exposure to diesel exhaust, TiO2 particles, CB,
30 or toner particles, it was reported that only the surface area of retained particles in the lung
31 showed a reasonable concentration-response relationship relative to tumor incidence and that
December 1994 10-20 DRAFT-DO NOT QUOTE OR CITE
-------
1 particle overload (retained mass or volume of particles) alone may not be the determining
2 factor in lung tumor formation. In this respect, it was shown that particles lacking adsorbed
3 organics (pure CB or TiO2 particles) and diesel exhaust particles exhibited a similar
4 relationship between particle surface area and tumor incidence. The investigators
5 hypothesized a tumorigenic effect would probably require that a "critical" surface area of
6 retained particles be attained for the manifestation of any mechanisms of tumorigenicity.
7 The possibility of a particle effect in the tumorigenic response has also been
8 demonstrated by Heinrich (1990) hi which female Wistar rats (72 per group) were exposed to
9 Printex 90 CB particles for 10 mo followed by a 20-mo exposure-free observation period or
10 for 20 mo followed by a 10-mo exposure-free observation period. A particle concentration
11 of 6.09 mg/m3 was used in both protocols. The Printex 90 particles had an extremely low
12 organic content («1,000-fold less than that of diesel exhaust particles). The tumor rates for
13 the 10- and 20-mo exposure durations were 17% (14% malignant) and 8% (all malignant),
14 respectively. Although the lower tumor incidence for the longer exposure period was not
15 consistent, the results demonstrate that the tumor incidences for CB particles with an organic
16 content 1,000-fold less than diesel exhaust particles are equivalent to those reported for diesel
17 exhaust exposures. The fact that these particles were able to exert a significant tumorigenic
18 response implicates the carbon core of diesel exhaust particles as possible tumor initiators in
19 diesel exhaust-induced carcinogenicity at high particle concentrations.
20 Preliminary findings regarding the importance of the particle-associated organics in
21 the pulmonary carcinogenicity of inhaled diesel exhaust in rats were reported by Mauderly
22 et al. (1991). In this long-term exposure study, rats were exposed 16 h/day, 5 days/week for
23 24 mo to whole diesel exhaust or CB (free of adsorbed organics) at particle concentrations of
24 2.5 or 6.5 mg/m3. Controls were exposed to clean air. Lung weights were increased in rats
25 exposed to the highest concentrations of both diesel exhaust or CB but were slightly higher
26 for the diesel exhaust group. The lung burdens of particulate matter were significantly
27 greater for the diesel exhaust-exposed rats at 18 and 23 mo. A substantial transfer of
28 particles from the lungs to lung-associated lymph nodes was observed, but no difference was
29 noted between the diesel exhaust and CB exposure groups. Inflammation and cytotoxicity
30 detected in lavage fluid was greater for diesel exhaust-exposed rats, but the difference was
31 proportional to the higher lung burden of retained particles noted for these animals.
December 1994 10-21 DRAFT-DO NOT QUOTE OR CITE
-------
1 Preliminary data based on approximately 100 male and 100 female rats indicated that the
2 numbers of lung tumors observed grossly at necropsy were nearly identical for the diesel
3 exhaust and CB exposure groups. Tumor type observed included squamous cysts, squamous
4 cell carcinomas, papillary adenocarcinomas, tubular adenocarcinomas, and solid carcinomas.
5 The growth of tumors transplanted into athymic mice has also been similar for diesel exhaust
6 and CB exposures, 74 and 73%, respectively. In summary, these preliminary observations
7 suggest that no difference exists hi the type or incidence of lung tumors hi rats following
8 long-term exposure to diesel exhaust or CB, and that the particle-associated organics may not
9 significantly involved in the pulmonary carcinogenicity of diesel exhaust hi rats.
10 The carcinogenic potential of many PAHs is well-documented, and, therefore, the
11 potential involvement of PAHs in diesel exhaust-induced carcinogenesis must be considered.
12 However, the recent reports by Heinrich (1990), Mauderly et al. (1991), and Nikula et al.
13 (1994) provide data that call into question the importance of PAHs in diesel exhaust-induced
14 carcinogenesis in these types of experiments that use exceptionally high particle
15 concentrations. Furthermore, a recent report (Bond et al., 1990a; see Section 10.4 for a
16 more detailed discussion of this work) reported that DNA adduct levels were similar in
17 Type II cells of rats exposed either to diesel exhaust or carbon black particles. Although
18 speculative at this time, the information in these studies suggest that PAHs may not be
19 instrumental in diesel exhaust-induced carcingenicity.
20
21
22 10.3 POTENTIAL INVOLVEMENT OF PULMONARY LEUKOCYTES
23 IN THE DEVELOPMENT OF LUNG TUMORS
24 Phagocytic leukocytes have been shown by numerous investigators to be toxic to
25 tumor cells in vivo, and increasing evidence suggests that cells of the mononuclear phagocyte
26 series hi particular may be of pivotal importance hi providing protection against malignancy
27 in situ. This protective function may, at least hi part, result from then- ability to produce a
28 tumor necrosis factor (TNF) (Urban et al., 1986). Whether the tumor surveillance and
29 tumoricidal activities of AMs (Hengst et al., 1978; Sone et al., 1983; Sone, 1986;
30 Kan-Mitchell et al., 1985) are compromised or otherwise modified when they are engorged
31 with even relatively benign particles has not been experimentally evaluated. The possibility
December 1994 10-22 DRAFT-DO NOT QUOTE OR CITE
-------
1 remains that diesel and other types of particles at high lung burdens result in decreases in
2 natural killer (NK) cell functional activities in providing defense against tumor formation
3 either by direct particle-cell interactions or by altering the ability of AMs to influence NK
4 cell-mediated host defense against metastatic tumor cells (Sone, 1986). These cells are
5 subpopulations of lymphocytes that possess spontaneous cytolytic activity toward neoplastic
6 cells but not toward normal cells. Moreover, the tumoricidal function of cytotoxic
7 T lymphocytes (Sone, 1986) may be directly or indirectly compromised by the presence of
8 high lung burdens of particles hi the lungs.
9 Phagocytes from a variety of species produce elevated levels of oxidant reactants in
10 response to challenges such as phagocytic stimuli, with the physicochemical characteristics of
11 a phagocytized particle being a major factor in determining the magnitude of the oxidant-
12 producing response. Hatch and co-workers (1980) have demonstrated that interactions of
13 guinea pig AMs with a wide variety of particles including silica, metal oxide-coated fly ash,
14 polymethylmethacrylate beads, chrysotile asbestos, fugitive dusts, polybead carboxylate
15 microspheres, glass and latex beads, uncoated fly ash, and fiberglass increase the production
16 of reactive oxygen species. Similar findings have been reported by numerous investigators
17 for human, rabbit, mouse, and guinea pig AMs (Drath and Karnovsky, 1975; Allen and
18 Loose, 1976; Beall et al., 1977; Lowrie and Aber, 1977; Miles et al., 1977; Rister and
19 Baehner, 1977; Hoidal et al., 1978). As well, polymorphonuclear leukocytes (PMNs) are
20 also known to increase production of superoxide radical, hydrogen peroxide, and hydroxyl
21 radical in response to membrane-reactive agents and particles (Goldstein et al., 1975; Weiss
22 et al., 1978; Root and Metcalf, 1977). Sagai et al. (1993) reported further that diesel
23 exhaust particles were able to produce superoxide and hydroxyl radicals in vitro, without
24 biological activation. Methanol washed particles, in this study, were much less toxic
25 following intratracheal instillation in the mouse, indicating that the active components were
26 extractable with organic solvents. It is well recognized that the deposition of particles in the
27 lung can result in the efflux of PMNs from the vascular compartment into the alveolar space
28 compartment in addition to expanding the AM population size. Following acute exposures,
29 the influx of the PMNs is transient, lasting only a few days (Adamson and Bowden, 1978;
30 Bowden and Adamson, 1978; Lehnert et al., 1988). Strom (1984) has reported that PMNs
31 become abnormally abundant following chronic exposures to paniculate diesel exhaust.
December 1994 10-23 DRAFT-DO NOT QUOTE OR CITE
-------
1 In the study by Strom (1984), the numbers of PMNs lavaged from the lungs of diesel-
2 exposed rats generally increased with increasing exposure duration and inhaled exposure
3 mass concentration. Strom (1984) also found that PMNs in diesel-exposed lungs remained
4 persistently elevated for at least 4 mo after cessation of exposure, a potential mechanism for
5 which may be related to an ongoing release of previously phagocytized particles by AMs that
6 engulfed them shortly after deposition. Evidence in support of this possibility has been
7 obtained by Lehnert et al. (1989) in a study in which rats were intratracheally instilled with
8 0.85, 1.06, or 3.6 mg of polystyrene particles. The PMNs were not found to be abnormally
9 abundant during the clearance of the two lower lung burdens, but they did become
10 progressively elevated in the lungs of the animals in which alveolar-phase clearance was
11 impaired. Moreover, the particle burdens in the PMNs became progressively greater over
12 time. Such findings are consistent with an ongoing particle relapse process, given the
13 relatively short lifespan of PMNs. As previously indicated, lung tumors develop in the rat at
14 lesser lung burdens of diesel exhaust particles than with a particle like TiO2.
15 Polymorphonuclear leukocytes characteristically are increased abnormally in the lung by
16 diesel exhaust exposure, but their presence in the lungs does not appear to be excessive
17 following the pulmonary deposition of even high lung burdens of TiO^ (Strom, 1984;
18 Lee et al., 1986). Thus, the generation of reactive oxygen species by both AMs and PMNs
19 should be considered as one potential factor of what probably is a multistep process that
20 culminates in the development of lung tumors in response to chronic deposition of diesel
21 exhaust particles.
22 As previously indicated, the production of oxygen species may afford protection
23 against emerging tumor cells by killing the cells, while under other conditions the production
24 of reactive oxygen products conceivably may actually contribute to the development of
25 neoplastic cells. The potential involvement of AMs and PMNs in the development of lung
26 tumors in laboratory rats administered high lung burdens of diesel particles (Mauderly et al.,
27 1987) or having inhaled particles that are generally considered to have low to no cytotoxic
28 potential (e.g., TiO2 [Lee et al., 1986]) over a prolonged period of time may be related to
29 the ability of the lung-free cells to produce reactive oxygen metabolites during phagocytic
30 oxidative metabolism (Hatch et al., 1980). Whereas products of phagocytic oxidative
31 metabolism, including superoxide anion, hydrogen peroxide, and hydroxyl radical can kill
December 1994 10-24 DRAFT-DO NOT QUOTE OR CITE
-------
1 tumor cells (Klebanoff and Clark, 1978), and the reactive oxygen species can peroxidize
2 lipids to produce cytotoxic metabolites such as malonyldialdehyde, some products of
3 oxidative metabolism apparently can also interact with DNA to produce mutations. Along
4 this line, Weitzman and Stossel (1981) found that human peripheral leukocytes were
5 mutagenic in the Ames assay. This mutagenic activity was related to PMNs and blood
6 monocytes; blood lymphocytes alone were not mutagenic. These investigators speculated
7 that the mutagenic activity of the phagocytes was a result of their ability to produce reactive
8 oxygen metabolites, inasmuch as blood leukocytes from a patient with chronic granulomatous
9 diseases, a disease in which neutrophils have a defect in the NADPH oxidase generating
10 system (Klebanoff and Clark, 1978), were less effective in producing mutations than were
11 normal leukocytes. Of related significance in terms of oxygen species being able to cause
12 genetic disturbances, Phillips et al. (1984) demonstrated that the incubation of Chinese
13 hamster ovary cells with xanthine plus xanthine oxidase (a system for enzymatically
14 generating active oxygen species) resulted in genetic damage hallmarked by extensive
15 chromosomal breakage and sister-chromatid exchange and produced an increase in the
16 frequency of thioguanidine-resistant cells (HGPRT test). Aside from interactions of oxygen
17 species with DNA, increasing evidence also points to an important role of phagocyte-derived
18 oxidants and/or oxidant products in the metabolic activation of procarcinogens to their
19 ultimate carcinogenic form (Kensler et al., 1987).
20 Another characteristic of AMs and PMNs that may contribute to the pathological
21 process is the production of a variety of regulatory and cytotoxic factors. Alveolar
22 macrophages can be stimulated to release several effector molecules or cytokines capable of
23 numerous regulatory functions of other lung cells, including rates of proliferation (Bitterman
24 et al., 1983; Jordana et al., 1988; Denholm and Phan, 1989). The involvement of these
25 factors in diesel exhaust pathology is only beginning to be explored. Release of tumor
26 necrosis factor (TNF) and interleukin-1 (IL-1) are correlated with increasing macrophage
27 particle load (Driscoll and Maurer, 1991). Both share a number of proinflammatory
28 activities, including neutrophil activation and chemotactic effects. Tumor necrosis factor-
29 associated phagocytic cell attraction is mediated through stimulation of lung macrophages;
30 fibroblasts or epithelial cells to secrete the ecosaniods MlP-la, MIP-2, and MCP; peptides
31 that are highly chemotactic to neutrophils; and monocytes (Driscoll et al., 1994). Tumor
December 1994 10-25 DRAFT-DO NOT QUOTE OR CITE
-------
1 necrosis factor has also been implicated in the induction of adhesion molecule expression
2 (Bevilacqua et al., 1989). Finally, TNF also stimulates phagocytic cells to release reactive
3 oxygen species and lysosomal enzymes (Klebanoff et al., 1986). Both SiO2 and TiO2
4 induced a persistent increase in fibronectin release by lung cells that was consistently
5 correlated with the development of pulmonary fibrosis (Driscoll and Maurer, 1991).
6 Transforming growth factor, TGF-51, present in macrophages and fibroblasts from silica-
7 exposed animals also appears to play a pathogenic role in particle-induced mesenchymal and
8 epithelial lesions (Williams et al., 1993).
9 A final characteristic of AMs and PMNs that may contribute to the pathologic process
10 leading to lung tumor development following diesel exhaust particle deposition is that these
11 phagocytes are known to release a variety of potentially destructive hydrolytic enzymes, a
12 process known to occur simultaneously with the phagocytosis of particles (Sandusky et al.,
13 1977). The essentially continual release of such enzymes during chronic particle deposition
14 and phagocytosis in the lung may be detrimental to the alveolar epithelium, especially to
15 Type I cells. Evans et al. (1986) showed that injury to Type I cells is followed shortly
16 thereafter by a proliferation of Type II cells. Type II cell hyperplasia is a generally common
17 feature observed in the lungs of animals that have received high lung burdens of various
18 types of particles, including unreactive polystyrene microspheres. Exaggerated proliferation
19 as a repair or defensive response to diesel deposition may have the effect of amplifying the
20 likelihood of neoplastic transformation in the presence of carcinogens beyond that which
21 would normally occur with lower rates of proliferation, assuming an increase in the cell
22 cycling of target cells and the probability of a neoplastic-associated genomic disturbance.
23 The proliferative response of Type II cells following the deposition of diesel exhaust
24 particles or other types of particles, however, has yet to be directly related to a Type I cell
25 destruction by proteolytic enzymes released by lung phagocytes or to a direct action of
26 particles on the proliferation kinetics of the Type II cells. The production of reactive oxygen
27 species or products therefrom could also be involved in the process. Whatever the stimulus,
28 it remains possible that the lung's AM population may play a role aside from any
29 responsibility for Type I cell damage. Alveolar macrophages have the ability to release
30 several other effector molecules or cytokines that can regulate numerous functions of other
31 lung cells, including their rates of proliferation (Bitterman et al., 1983; Jordana et al., 1988;
December 1994 10-26 DRAFT-DO NOT QUOTE OR CITE
-------
1 Denholm and Phan, 1989). The AM-derived mediators that may stimulate Type II cell
2 hyperplasia following particle deposition in the lung, however, remain to be identified, if in
3 fact the AMs play a regulatory role in the Type II cell proliferative response.
4
5
6 10.4 MOLECULAR DOSIMETRY CONSIDERATIONS
7 An important component in understanding diesel exhaust-induced carcinogenicity is
8 understanding the molecular basis for such an effect. As previously described, considerable
9 data exist regarding DNA adducts resulting from metabolites of organic compounds
10 associated with diesel exhaust particles. For the most part, however, these studies evaluated
11 specific organic compounds or used diesel paniculate matter to which had been adsorbed
12 unusually high levels of the organic compounds. This section focuses on those studies
13 evaluating DNA adduct formation in the lungs of animals exposed to diesel exhaust and
14 evaluates dosimetric parameters relating to adduct formation.
15 DNA adduct formation in the lungs of animals subjected to long-term exposure to
16 whole diesel exhaust has been described by Wong et al. (1986). Using tissues from animals
17 of the Mauderly et al. (1987) study, these investigators reported an increase in DNA adduct
18 formation in male and female F344 rats exposed to whole diesel exhaust (7.1 mg of
19 particles/m3) for 7 h/day, 5 days/week for up to 30 mo. P32 postlabeling was applied to
20 DNA that was extracted from six control and six exhaust-exposed rats (males and females).
21 Characterization of the adducts and identification of the exhaust components responsible for
22 their formation were not within the scope of the study. The lungs of exhaust-exposed rats
23 were darkly pigmented and contained diesel-particle-laden macrophages. Aggregates of these
24 macrophages were frequently associated with alveolar wall fibrosis, bronchiolar metaplasia,
25 and, occasionally, squamous metaplasia. Lungs from control rats were not darkly pigmented
26 and had relatively unaltered airway and structure. Autoradiographic analysis revealed
27 elevated levels of DNA adducts in the exhaust-exposed rats. The authors indicated that
28 quantitative and qualitative data regarding DNA adducts resulting from diesel exhaust
29 exposure may be useful for extrapolation to potential effects in humans.
30 A study by Bond et al. (1989) addressed several key topics regarding the role of DNA
31 adducts in the pulmonary carcinogenicity of diesel exhaust. Using groups of rats exposed to
December 1994 10-27 DRAFT-DO NOT QUOTE OR CITE
-------
1 whole diesel exhaust at particle concentrations of 0, 0.35, 3.5, 7.0, or 10.0 mg/m3 for
2 12 weeks, the relationship between DNA adduct levels and exposure concentration was
3 examined. The data for the exposure levels employed indicated that DNA adduct formation
4 (about 14 adducts per 109 bases) was similar across all exposure concentrations and was
5 approximately twice that of the sham-exposed group. The fact that DNA adduct formation
6 was independent of exposure concentration may be explained, in part, by previously reported
7 (Bond and Mauderly, 1984) data showing that metabolism of organics associated with diesel
8 exhaust by the isolated perfused rat lung could be saturated at high concentrations, thereby
9 limiting the production of metabolites required for the formation of DNA adducts.
10 The time course for DNA adduct formation was also examined by Bond et al. (1989).
11 Over a 12-week period of exposure to diesel exhaust (7 mg/m3), lung DNA adducts were
12 found to slowly accumulate. The highest adduct level was reached at 12 weeks, followed by
13 a decline to control level by 4 weeks postexposure. Throughout the exposure period, lung
14 DNA adducts remained constant and at a lower level in sham-exposed rats. The investigators
15 suggested that the rapid repair of adducts relative to their formation might result in a steady -
16 state level of DNA adducts during long-term exposure.
17 A dosimetry study examined the distribution of DNA adducts in the respiratory tract
18 to determine if increased DNA adduct formation occurred in regions of the lung where diesel
19 exhaust-induced tumors are formed (Bond et al., 1988). For this study, rats were exposed
20 for 12 weeks to diesel exhaust at a particle concentration of 10 mg/m3. The DNA adduct
21 levels were highest in peripheral tissue, which is the same region in which tumors occurred
22 in rats in long-term exposure studies (Mauderly et al., 1987). Although these findings
23 suggest that DNA adduct formation and tumor formation are related, the data do not prove
24 the association.
25 The previous studies provided data regarding the role of DNA adducts in the
26 pulmonary carcinogenesis of diesel exhaust but were not designed to provide insight into
27 possible target cells. An additional molecular dosimetry study by Bond et al. (1990a)
28 addressed this topic and also compared the effects of diesel exhaust particles with CB
29 particles that were virtually free of the adsorbed organics found on diesel exhaust particles.
30 In this study, rats were exposed to whole diesel exhaust (6.2 mg/m3), CB particles
31 (6.2 mg/m3), or clean air, 16 h/day, 5 days/week for 12 weeks. Relative to clean-air
December 1994 10-28 DRAFT-DO NOT QUOTE OR CITE
-------
1 controls, a significant increase in the total DNA adduct level in Type II cells was noted for
2 rats exposed to diesel exhaust and CB. The exposure to CB and diesel exhaust resulted in an
3 approximate fourfold increase in adduct level compared with controls. However, the
4 investigators noted that there was a large region of unresolved adducts in the chromatograms
5 from diesel-exhaust-exposed rats and that the total adduct level in these animals may be
6 underestimated. Whether the small amount (= 0.04%) of extractable organics on the CB
7 particles was responsible for the observed DNA adduct formation or the adducts were the
8 result of inflammatory responses to the particles was not determined. This study does,
9 however, demonstrate that Type II cells are possible targets for diesel exhaust exposure.
10 Although the actual mechanisms of diesel particle induced cancer induction remain to
11 be elucidated, a general scheme can be hypothesized. First of all, stimulation of chemotaxis
12 and cell adhesion results in aggregation of particle laden macrophages. The oxidants,
13 lysosomal enzymes, eluted organics, and other cytoxic agents released by these macrophages
14 are likely to be concentrated in the immediate region. In fact, alveolar epithelial lesions
15 were reported adjacent to these aggregations in diesel exhaust-exposed rats (Mauderly et al.,
16 1987). In mice, which appear to be less susceptible to the pathological effects of particles,
17 the macrophages remain more disperse (Mauderly, 1994). The role of these various factors
18 in particle-induced cancer induction is still subject to debate. However, it is reasonable to
19 assume that the PAHs will have tumor-initiating effects through direct action on DNA. The
20 cytokines, on the other hand, may cause cell proliferation creating a dividing cell population
21 that may be initiated through exogenous or endogenous process agents. It should be
22 recognized that proliferative effects may include interfering with normal cell death processes.
23
24
25 10.5 SUMMARY OF METABOLISM AND MECHANISM OF ACTION
26 OF CARCINOGENIC COMPONENTS OF DIESEL EXHAUST
27 Currently, a determination of whether the carcinogenicity of diesel exhaust is the
28 result of genetic or epigenetic mechanisms or a combination of both is uncertain (Bond et al.,
29 1989). Based on current data, diesel-exhaust-induced carcinogenesis appears to involve an
30 initiation-promotion mechanism.
December 1994 !0-29 DRAFT-DO NOT QUOTE OR CITE
-------
1 The genetic mechanism is supported by data showing the formation of DNA adducts
2 in exposed animals and by the known carcinogenic and mutagenic potential of many of the
3 compounds in diesel exhaust. Several studies affifm the bioavailability from inhaled diesel
4 exhaust particles of compounds such as B[a]P and 1-NP, which are known to be carcinogenic
5 or mutagenic. Furthermore, that xenobiolics may undergo biotransformation to reactive
6 intermediates following their entry into the body via inhalation of diesel exhaust particles has
7 been demonstrated for B[a]P and various nitroarertes. Results from the metabo-
8 lism/disposition studies using carbon particles to which organics have been experimentally
9 adsorbed must be interpreted with caution, however. The concentration of organics on these
10 particles is probably much greater than the monomolecttlar or bimolecular layer on actual
11 diesel exhaust particles and, therefore, might facilitate desorption of the organics from these
12 experimentally prepared particles.
13 It is generally accepted that one of the underlying mechanisms of carcinogenesis
14 involves the formation of covalent adducts with DNA, resulting in the alteration of cellular
15 genetic information. Several reports have provided data indicating that such adducts are
16 formed in animals following administration of these compounds and after long-term exposure
17 to diesel exhaust. The premise that DNA adduct formation plays a role in diesel exhaust-
18 induced carcinogertesis is substantiated by several findings, including an increase in DNA
19 adducts in the same pulmonary regions where tumors occur, and higher DNA adduct levels
20 in species known to be susceptible to diesel-exhaust-induced tumors. However, the lack of
21 an exposure-response for DNA adduct formation as demonstrated by the molecular dosimetry
22 studies reported by Bond et al. (1990b) suggest the involvement of additional mechanisms.
23 It is clear that an understanding of diesel-exhaust-induced carcinogenicity will require a more
24 complete knowledge of the metabolism and kinetic parameters that relate to adduct formation
25 and the involvement of genetic repair processes.
26 The presence of a promotional mechanism is supported by the fact that carbon
27 particles per se cause inflammatory responses and increased epithelial cell proliferation and
28 that alveolar macrophage function may be compromised under conditions of particle
29 overload. Studies using CB particles having very low levels of adsorbed organics have been
30 shown to increase inflammatory responses in the lung and to increase adduct formation.
31 Furthermore, recent studies have shown tumor rates resulting from exposures to nearly
December 1994 10-30 DRAFT-DO NOT QUOTE OR CITE
-------
1 organic-free CB particles to be similar to those observed for diesel exhaust exposures, thus
2 providing some evidence for epigenetic mechanisms.
3 The development of lung tumors in experimental laboratory animals following chronic
4 exposures to diesel exhaust occurs under conditions in which AM-mediated particle clearance
5 from the lung is compromised. As previously noted, tumors have also been found to develop
6 with other types of particles when this clearance mechanism is diminished. Thus, reductions
7 in the functional activity of the lung's alveolar macrophage population in the clearance
8 process generally appear to be intimately related to the carcinogenic response to high lung
9 burdens of particles. Findings that tumors develop in the lungs of laboratory rats at lesser
10 lung mass or volume burdens of diesel particles than with a substance such as TiO2 suggest
11 that the carcinogenic response, however, is not exclusively related to an overabundance of
12 particles in the lungs per se. That the organic components on diesel particles, many of
13 which have demonstrated carcinogenic activity, may be involved in the development of lung
14 tumors is a reasonable hypothesis. The lung's AMs, which phagocytize deposited diesel
15 particles, probably participate in the gradual in situ extraction and metabolism of
16 procarcinogens associated with the diesel particles. Additionally, the normal tumoricidal
17 activities of the AMs may be compromised upon interaction with excessive numbers of diesel
18 particles, and diesel particle-macrophage interactions could lead to the generation of reactive
19 oxygen species that have been shown to be at least mutagenic. Another hypothesis is that the
20 organics are unnecessary for the diesel exhaust-induced tumorigenic effect and that it is the
21 carbon particle alone that is responsible. Both diesel exhaust particles and carbon black
22 particles have greater surface areas and are more effective at tumor induction than are TiO2
23 particles. Because carbon black particles are essentially devoid of adsorbed organics, the
24 tumorigenic response observed for both of these may be due the effects of the particles alone.
25 The resolution of the relative contributions of chemical carcinogens and particle overload to
26 the tumors occurring in experimental animals exposed to diesel exhaust needs immediate
27 research attention.
28 Caution must be exercised in extrapolating observations made in animal models to
29 humans. Processes and potential mechanisms discussed herein have largely been derived
30 from animal data. Further research is required to determine how the response of human
31 AMs to paniculate diesel exhaust compares with that of AMs in experimental animals at
December 1994 10-31 DRAFT-DO NOT QUOTE OR CITE
-------
1 particle concentrations typical of human exposure scenarios. Most important, valid
2 dosimetry for humans requires the elucidation of the underlying mechanisms involved in the
3 development of lung tumors following chronic exposure to whole diesel exhaust.
December 1994 10-32 DRAFT-DO NOT QUOTE OR CITE
-------
1 REFERENCES
2 Adamson, I. Y. R.; Bowden, D. H. (1978) Adaptive responses of the pulmonary macrophagic system to carbon:
3 II. morphologic studies. Lab. Invest. 38: 430-438.
4
5 Allen, R. C,; Loose, L. D. (1976) Phagocytic activation of a luminol-dependenl chemiluminescence in rabbit
6 alveolar and peritoneal macrophages. Biochem. Biophys. Res. Commun. 69: 245-252.
7
8 Ball, L. M.; King, L. C. (1985) Metabolism, mutagenicity, and activation of 1-nitropyrene in vivo and in vitro.
9 Environ. Int. 11: 355-361.
10
11 Beall, G. D.; Repine, J. E.; Hoidal, J. R.; Rasp, F. L. (1977) Chemiluminescence by human alveolar
12 macrophages: stimulation with heat-killed bacteria or phorbol myristate acetate. Infect. Immun.
13 17: 117-120.
14
15 Bevilacqua, M. P.; Stengelin, S.; Gimbrone, M. A.; Seed, B. (1989) Endothelial leukocyte adhesion molecule
16 1: an inducible receptor for neutrophils related to complement regulatory proteins and lectins. Science
17 (Washington, DC) 243: 1160-1165.
18
19 Bitterman, P. B.; Adelberg, S.; Crystal, R. G. (1983) Mechanisms of pulmonary fibrosis: spontaneous release of
20 the alveolar macrophage-derived growth factor in the interstitial lung disorders. J. Clin. Invest.
21 72: 1801-1813.
22
23 Bond, J. A.; Mauderly, J. L. (1984) Metabolism and macromolecular covalent binding of [14C]-1-nitropyrene in
24 isolate perfused and ventilated rat lungs. Cancer Res. 44: 3924-3929.
25
26 Bond, J. A.; Mitchell, C. E.; Li, A. P. (1983) Metabolism and macromolecular covalent binding of
27 benzo[a]pyrene in cultured Fischer-344 rat lung type II epithelial cells. Biochem. Pharmacol.
28 32: 3771-3776.
29
30 Bond, J. A.; Butler, M. M.; Medinsky, M. A.; Muggenburg, B. A.; McClellan, R. O. (1984) Dog pulmonary
31 macrophage metabolism of free and particle-associated [14C]benzo[a]pyrene. J. Toxicol. Environ. Health
32 14: 181-189.
33
34 Bond, J. A.; Sun, J. D.; Medinsky, M. A.; Jones, R. K.; Yeh, H. C. (1986) Deposition, metabolism, and
35 excretion of l-[14C]nitropyrene and l-[14CJnitropyrene coated on diesel exhaust particles as influenced by
36 exposure concentration. Toxicol. Appl. Pharmacol. 85: 102-117.
37
38 Bond, J. A.; Wolff, R. K.; Harkema, J. R.; Mauderly, J. L.; Henderson, R. F.; Griffith, W. C.; McClellan,
39 R. O. (1988) Distribution of DNA adducts in the respiratory tract of rats exposed to diesel exhaust.
40 Toxicol. Appl. Pharmacol. 96: 336-346.
41
42 Bond, J. A.; Harkema, J. R.; Henderson, R. F.; Mauderly, J. L.; McClellan, R. 0.; Wolff, R. K. (1989)
43 Molecular dosimetry of inhaled diesel exhaust. In: Mohr, U.; Bates, D. V.; Dungworth, D. L.; Lee,
44 P. N.; McClellan, R. O.; Roe, F. J. C., eds. Assessment of inhalation hazards: integration and
45 extrapolation using diverse data: [papers from the 2nd international inhalation symposium]; 1989;
46 Hannover, FRG. Berlin, Federal Republic of Germany: Springer-Verlag; pp. 315-324.
47
48 Bond, J. A.; Johnson, N. F.; Snipes, M. B.; Mauderly, J. L. (1990a) DNA adduct formation in rat alveolar
49 type II cells: cells potentially at risk for inhaled diesel exhaust. Environ. Mol. Mutagen. 16: 64-69.
51 Bond, J. A.; Mauderly, J. L.; Wolff, R. K. (1990b) Concentration- and time-dependent formation of DNA
52 adducts in lungs of rats exposed to diesel exhaust. Toxicology 60: 127-135.
53
December 1994 10-33 DRAFT-DO NOT QUOTE OR CITE
-------
1 Bowden, D. H.; Adamson, I. Y. R. (1978) Adaptive responses of the pulmonary macrophagic system to carbon:
2 I. kinetic studies. Lab. Invest. 38: 422-429.
3
4 Boyd, M. R. (1984) Metabolic activation and lung toxicity: a basis for cell-selective pulmonary damage by
5 foreign chemicals. Environ. Health Perspect. 55: 47-51.
6
7 Boyland, E. (1980) The history and future of chemical carcinogenesis. Br. Med. Bull. 36: 5-10.
8
9 Cantrell, E. T.; Tyrer, H. W.; Peirano, W. B.; Danner, R. M. (1980) Benzo(a)pyrene metabolism in mice
10 exposed to diesel exhaust: II. metabolism and excretion. In: Pepelko, W. E.; Danner, R. M.; Clarke,
11 N. A., eds. Health effects of diesel engine emissions: proceedings of an international symposium, v. 1;
12 December 1979; Cincinnati, OH. Cincinnati, OH: U.S. Environmnetal Protection Agency, Health Effects
13 Research Laboratory; pp. 520-531. EPA report no. EPA-600/9-80-057a. Available from: NTIS,
14 Springfield, VA; PB81-173809.
15
16 Chen, K.-C.; Vostal, J. J. (1982) Aryl hydrocarbon hydroxylase induction in rat pulmonary alveolar
17 macrophages (PAM) by diesel particulates or their extracts. Pharmacologist 24: 151.
18
19 Conney, A. K. (1982) Induction of microsomal enzymes by foreign chemicals and carcinogenesis by polycyclic
20 aromatic hydrocarbons: G. H. A. Clowes memorial lecture. Cancer Res. 42: 4875-4917.
21
22 Crapo, J. D.; Young, S. L.; Fram, E. K.; Pinkerton, K. E.; Barry, B. E.; Crapo, R. O. (1983) Morphometric
23 characteristics of cells in the alveolar region of mammalian lungs. Am. Rev. Respir. Dis. 128: S42-S46.
24
25 Delclos, K. B.; Miller, D. W.; Lay, J. O., Jr.; Casciano, D. A.; Walker, R. P.; Fu, P. P.; Kadlubar, F. F.
26 (1987a) Identification of C8-modified deoxyinosine and N2- and C8-modified deoxyguanosine as major
27 products of the in vitro reaction of N-hydroxy-6-aminochrysene with DNA and the formation of these
28 adducts in isolated rat hepatocytes treated with 6-nitrochrysene and 6-aminochrysene. Carcinogenesis
29 (London) 8: 1703-1709.
30
31 Delclos, K. B.; Walker, R. P.; Dooley, K. L.; Fu, P. P.; Kadlubar, F. F. (1987b) Carcinogen-DNA adduct
32 formation in the lungs and livers of preweanling CD-I male mice following administration of
33 [3H]-6-nitrochrysene, [3H]-6-aminochrysene, and [3H]-l,6-dinitropyrene. Cancer Res. 47: 6272-6277.
34
35 Delclos, K. B.; El-Bayoumy, K.; Hecht, S. S.; Walker, R. P.; Kadlubar, F. F. (1988) Metabolic activation of
36 6-aminochrysene and 6-nitrochrysene: a diol-epoxide of 6-aminochrysene as a probable ultimate
37 carcinogen in preweanling mice. In: King, C. M.; Romano, L. J.; Schuetzle, D., eds. Carcinogenic and
38 mutagenic responses to aromatic amines and nitroarenes: proceedings of the third international conference
39 on carcinogenic and mutagenic N-substituted aryl compounds; April 1987; Dearborn, MI. New York,
40 NY: Elsevier; pp. 103-106.
41
42 Denholm, E. M.; Phan, S. H. (1989) The effects of bleomycin on alveolar macrophage growth factor secretion.
43 Am. J. Pathol. 134: 355-363.
44
45 Djuric, Z.; Fifer, E. K.; Beland, F. A. (1985) Acetyl coenzyme A-dependent binding of carcinogenic and
46 mutagenic dinitropyrenes to DNA. Carcinogenesis (London) 6: 941-944.
47
48 Djuric, Z.; Fifer, E. K.; Yamazoe, Y.; Beland, F. A. (1988) DNA binding by 1-nitropyrene and
49 1,6-dinitropyrene in vitro and in vivo: effects of nitroreductase induction. Carcinogenesis (London)
50 9: 357-364.
51
52 Drath, D. B.; Karnovsky, M. L. (1975) Superoxide production by phagocytic leukocytes. J. Exp. Med.
53 141:257-262.
54
December 1994 10-34 DRAFT-DO NOT QUOTE OR CITE
-------
1 Driscoll, K. E.; Maurer, J. K. (1991) Cytokine and growth factor release by alveolar macrophages: potential
2 biomarkers of pulmonary toxicity. Toxicol. Pathol. 19: 398-405.
3
4 Driscoll, K. E.; Maurer, J. K.; Hassenbien, D.; Carter, J.; Janssen, Y. M. W.; Mossman, B. T.; Osier, M.;
5 Oberdorster, G. (1994) Contribution of macrophage-derived cytokines and cytokine networks to mineral
6 dust-induced lung inflammation. In: Mohr, U.; Dungworth, D. L.; Mauder, J. L.; Oberdorster, G., eds.
7 Toxic and carcinogenic effects of solid particles in the respiratory tract: [proceedings of the 4th
8 international inhalation symposium]; March 1993; Hannover, Germany. Washington, DC: International
9 Life Sciences Institute Press; pp. 177-189.
10
11 El-Bayoumy, K.; Hecht, S. S. (1983) Identification and mutagenicity of metabolites of 1-nitropyrene formed by
12 rat liver. Cancer Res. 43: 3132-3137.
13
14 El-Bayoumy, K.; Hecht, S. S. (1984) Identification of fra«j-l,2-dihydro-l,2-dihydroxy-6-nitrochryseneas a
15 major mutagenic metabolite of 6-nitrochrysene. Cancer Res. 44: 3408-3413.
16
17 El-Bayoumy, K.; Rivenson, A.; Johnson, B.; DiBello, J.; Little, P.; Hecht, S. S. (1988) Comparative
18 tumorigenicity of 1-nitropyrene, 1-nitrosopyrene, and 1-aminopyrene administered by gavage to
19 Sprague-Dawley rats. Cancer Res. 48: 4256-4260.
20
21 Evans, M. J.; Shami, S. G.; Martinez, L. A. (1986) Enhanced proliferation of pulmonary alveolar macrophages
22 after carbon instillation in mice depleted of blood monocytes by Strontium-89. Lab. Invest. 54: 154-159.
23
24 Fu, P. P.; von Tungeln, L. S.; Unruh, L. E.; Heflich, R. H. (1986) In vitro metabolism of 4-nitropyrene to a
25 new type of mutagenic metabolite. Pharmacologist 28: 120.
26
27 Garshick, E.; Schenker, M. B.; Munoz, A.; Segal, M.; Smith, T. J.; Woskie, S. R.; Hammond, S. K.; Speizer,
28 F. E. (1987) A case-control study of lung cancer and diesel exhaust exposure in railroad workers.
29 Am. Rev. Respir. Dis. 135: 1242-1248.
30
31 Glatt, H. R.; Oesch, F. (1976) Phenolic benzo(a)pyrene metabolites are mutagens. Mutat. Res. 36: 379-383.
32
33 Goldstein, I. M.; Roos, D.; Kaplan, H. B.; Weissman, G. (1975) Complement and immunoglobulins stimulate
34 superoxide production by human leukocytes independently of phagocytosis. J. Clin. Invest.
35 56: 1155-1163.
36
37 Harris, C. C. (1985) Future directions in the use of DNA adducts as internal dosimeters for monitoring human
38 exposure to environmental mutagens and carcinogens. Environ. Health Perspect. 62: 185-191.
40 Hatch, G. E.; Gardner, D. E.; Menzel, D. B. (1980) Stimulation of oxidant production in alveolar macrophages
41 by pollutant and latex particles. Environ. Res. 23: 121-136.
42
43 Heflich, R. H.; Djuric, Z.; Fifer, E. K.; Cerniglia, C. E.; Beland, F. A. (1986) Metabolism of dinitropyrenes to
44 DNA-binding derivatives in vitro and in vivo. In: Ishinishi, N.; Koizumi, A.; McClellan, R. O.;
45 Stober, W., eds. Carcinogenic and mutagenic effects of diesel engine exhaust: proceedings of the
46 international satellite symposium on lexicological effects of emissions from diesel engines; July; Tsukuba
47 Science City, Japan. Amsterdam, The Netherlands: Elsevier Science Publishers B. V.; pp. 185-197.
48 (Developments in toxicological and environmental science: v. 13).
49
50 Heinrich, U. (1990) Results of long-term inhalation exposure of rats to carbon black "Printex 90" [letter to
51 Dr- Lester D. Grant]. Presented at: U.S. Environmental Protection Agency peer review workshop on the
52 Health Assessment Document for Diesel Emissions; July; Research Triangle Park, NC.
December 1994 10_35 DRAFT-DO NOT QUOTE OR CITE
-------
1 Heinrich, U.; Muhle, H.; Takenaka, S.; Ernst, H.; Fuhst, R.; Mohr, U.; Pott, F.; Stober, W. (1986) Chronic
2 effects on the respiratory tract of hamsters, mice, and rats after long-term inhalation of high
3 concentrations of filtered and unfiltered diesel engine emissions. J. Appl. Toxicol. 6: 383-395.
4
5 Hibbs, J. B., Jr.; Chapman, H. A., Jr.; Weinberg, J. B. (1978) The macrophage as an antineoplastic surveillance
6 cell: biological perspectives. J. Reticuloendothel. Soc. 24: 549-570.
7
8 Hirose, M.; Lee, M.-S.; Wang, C. Y.; King, C. M. (1984) Induction of rat mammary gland tumors by
9 1-nitropyrene, a recently recognized environmental mutagen. Cancer Res. 44: 1158-1162.
10
11 Hoidal, J. R.; Repine, J. E.; Beall, G. D.; Rasp, F. L., Jr.; White, J. G. (1978) The effect of phorbol myristate
12 acetate on the metabolism and ultrastructure of human alveolar macrophages. Am. J. Pathol.
13 91: 469-482.
14
15 Howard, P. C.; Gerrard, J. A.; Milo, G. E.; Fu, P. P.; Beland, F. A.; Kadlubar, F. F. (1983) Transformation
16 of normal human skin fibroblasts by 1-nitropyrene and 6-nitrobenzo[a]pyrene. Carcinogenesis (London)
17 4: 353-355.
18
19 Howard, A. J.; Mitchell, C. E.; Dutcher, J. S.; Henderson, T. R.; McClellan, R. O. (1986) Binding of
20 nitropyrenes and benzo[a]pyrene to mouse lung deoxyribonucleic acid after pretreatment with inducing
21 agents. Biochem. Pharmacol. 35: 2129-2134.
22
23 Imaida, K.; Tay, L. K.; Lee, M.-S.; Wang, C. Y.; Ito, N.; King, C. M. (1988) Tumor induction by
24 nitropyrenes in the female CD rat. In: King, C. M.; Romano, L. J.; Schuetzle, D., eds. Carcinogenic
25 and mutagenic responses to aromatic amines and nitroarenes: proceedings of the third international
26 conference on carcinogenic and mutagenic W-substituted aryl compounds; April 1987; Dearborn, MI.
27 New York, NY: Elsevier; pp. 187-197.
28
29 International Agency for Research on Cancer. (1989) Diesel and gasoline engine exhausts and some nitroarenes.
30 Lyon, France: World Health Organization; pp. 189-373. (IARC monographs on the evaluation of
31 carcinogenic risks to humans: v. 46).
32
33 Iwai, K.; Udagawa, T.; Yamagishi, M.; Yamada, H. (1986) Long-term inhalation studies of diesel exhaust on
34 F344 SPF rats. Incidence of lung cancer and lymphoma. In: Ishinishi, N.; Koizumi, A.; McClellan,
35 R. O.; Stober, W., eds. Carcinogenic and mutagenic effects of diesel engine exhaust: proceedings of the
36 international satellite symposium on toxicological effects of emissions from diesel engines; July; Tsukuba
37 Science City, Japan. Amsterdam, Holland: Elsevier Science Publishers B. V.; pp. 349-360.
38 (Developments in toxicology and environmental science: v. 13).
39
40 Jerina, D. M.; Sayer, J. M.; Thakker, D. R.; Yagi, H.; Levin, W.; Wood, A. W.; Conney, A. H. (1980)
41 Carcinogenicity of polycyclic aromatic hydrocarbons: the bay-region theory. In: Pullman, B.; Ts'o,
42 P. O. P.; Gelboin, H., eds. Carcinogenesis: fundamental mechanisms and environmental effects.
43 Dordrecht, The Netherlands: D. Reidel; pp. 1-12. (Jerusalem symposia on quantum chemistry and
44 biochemistry: v. 13).
45
46 Jordana, M.; Richards, C.; Irving, L. B.; Gauldie, J. (1988) Spontaneous in vitro release of alveolar-macrophage
47 cytokines after the intratracheal instillation of bleomycin in rats: characterization and kinetic studies.
48 Am. Rev. Respir. Dis. 137: 1135-1140.
49
50 Kan-Mitchell, J.; Hengst, J. C. D.; Kempf, R. A.; Rothbart, R. K.; Simons, S. M.; Brooker, A. S.; Kortes,
51 V. L.; Mitchell, M. S. (1985) Cytotoxic activity of human pulmonary alveolar macrophages. Cancer Res.
52 45: 453-458.
53
December 1994 10-36 DRAFT-DO NOT QUOTE OR CITE
-------
1 Kawabata, Y.; Iwai, K.; Udagawa, T.; Tukagoshi, K.; Higuchi, K. (1986) Effects of diesel soot on unscheduled
2 DNA synthesis of tracheal epithelium and lung tumor formation. In: Ishinishi, N.; Koizumi, A.;
3 McClellan, R. O.; Stober, W., eds. Carcinogenic and mutagenic effects of diesel engine exhaust:
4 proceedings of the international satellite symposium on toxicological effects of emissions from diesel
5 engines; July; Tsukuba Science City, Japan. Amsterdam, The Netherlands: Elsevier Science Publishers
6 B. V.; pp. 213-222. (Developments in toxicologial and environmental science: v. 13).
7
8 Kensler, T. W.; Egner, P. A.; Moore, K. G.; Taffe, B. G.; Twerdok, L. E.; Trush, M. A. (1987) Role of
9 inflammatory cells in the metabolic activation of polycyclic aromatic hydrocarbons in mouse skin.
10 Toxicol. Appl. Pharmacol. 90: 337-346.
11
12 King, C. M. (1988) Metabolism and biological effects of nitropyrene and related compounds. Cambridge, MA:
13 Health Effects Institute; research report no. 16.
14
15 King, L. C.; Jackson, M.; Ball, L. M.; Lewtas, J. (1983) Binding of l-nitro[14C]pyrene to DNA and protein in
16 cultured lung macrophages and respiratory tissues. Cancer Lett. (Shannon, Irel.) 19: 241-246.
17
18 King, C. M.; Tay, L. K.; Lee, M.-S.; Imaida, K.; Wang, C. Y. (1986) Machanism of tumor induction by
19 dinitropyrenes in the female CD rat. In: Ishinishi, N.; Koizumi, A.; McClellan, R. O.; Stober, W., eds.
20 Carcinogenic and mutagenic effects of diesel engine exhaust: proceedings of the international satellite
21 symposium on toxicological effects of emissions from diesel engines; July; Tsukuba Science City, Japan.
22 Amsterdam, The Netherlands: Elsevier Science Publishers B. V.; pp. 279-290. (Developments in
23 toxicology and environmental science: v. 13).
24
25 King, L. C.; Kohan, M. J.; George, S. E.; Lewtas, J.; Claxton, L. D. (1990) Metabolism of 1-nitropyrene by
26 human, rat, and mouse intestinal flora: mutagenicity of isolated metabolites by direct analysis of HPLC
27 fractions with a microsuspension reverse mutation assay. J. Toxicol. Environ. Health 31: 179-192.
28
29 Klebanoff, S. J.; Clark, R. A. (1978) The neutrophil: function & clinical disorders. Amsterdam, The
30 Netherlands: Elsevier.
31
32 Klebanoff, S. J.; Vadas, M. A.; Harlan, J. M.; Sparks, L. H.; Gamble, J. R.; Agosti, J. M.; Waltersdorph,
33 A. M. (1986) Stimulation of neutrophils by tumor necrosis factor. J. Immunol. 136: 4220-4225.
34
35 Lee, K. P.; Henry, N. W., Ill; Trochimowicz, H. J.; Reinhardt, C. F. (1986) Pulmonary response to impaired
36 lung clearance in rats following excessive TiO2 dust deposition. Environ. Res. 41: 144-167.
37
38 Lee, K. P.; Ulrich, C. E.; Geil, R. G.; Trochimowicz, H. J. (1988) Effects of inhaled chromium dioxide dust on
39 rats exposed for two years. Fundam. Appl. Toxicol. 10: 125-145.
40
41 Lehnert, B. E.; Valdez, Y. E.; Bomalaski, S. H. (1988) Analyses of particles in the lung free cell,
42 tracheobronchial lymph nodal, and pleural space compartments following their deposition in the lung as
43 related to lung clearance mechanisms. Ann. Occup. Hyg. 32: 125-140.
44
45 Lehnert, B. E.; Ortiz, J. B.; London, J. E.; Cline, A. F. (1989) Migratory behavior of alveolar macrophages
46 during clearance of light to heavy lung burdens of particles [abstract]. Am. Rev. Respir. Dis.
47 139(suppl.): A245.
48
49 Leung, H.-W.; Henderson, R. F.; Bond, J. A.; Mauderly, J. L.; McClellan, R. O. (1988) Studies on the ability
50 of rat lung and liver microsomes to facilitate transfer and metabolism of benzo[a]pyrene from diesel
51 particles. Toxicology 51: 1-9.
52
53 Lowrie, D. B.; Aber, V. R. (1977) Superoxide production by rabbit pulmonary alveolar macrophages. Life Sci.
54 21: 1575-1583.
December 1994 10-37 DRAFT-DO NOT QUOTE OR CITE
-------
1 Maher, V. M.; Patton, J. D.; McCormick, J. J. (1988) Studies on the metabolism and biological effects of
2 nitropyrene and related nitro-polycyclic aromatic compounds in diploid human fibroblasts. Cambridge,
3 MA: Health Effects Institute; research report no. 17.
4
5 Manabe, Y.; Kinouchi, T.; Ohnishi, Y. (1985) Identification and quantification of highly mutagenic
6 nitroacetoxypyrenes and nitrohydroxypyrenes in diesel-exhaust particles. Mutat. Res. 158: 3-18
7
8 Manning, B. W.; Campbell, W. L.; Franklin, W.; Delclos, K. B.; Cerniglia, C. E. (1988) Metabolism of
9 6-nitrochrysene by intestinal microflora. Appl. Environ. Microbiol. 54: 197-203.
10
11 Mauderly, J. L.; Jones, R. K.; Griffith, W. C.; Henderson, R. F.; McClellan, R. O. (1987) Diesel exhaust is a
12 pulmonary carcinogen in rats exposed chronically by inhalation. Fundam. Appl. Toxicol. 9: 208-221.
13
14 Mauderly, J. L.; Snipes, M. B.; Barr, E. B.; Bechtold, W. E.; Henderson, R. F.; Mitchell, C. E.; Nikula,
15 K. J.; Thomassen, D. G. (1991) Influence of particle-associated organic compounds on carcinogenicity of
16 diesel exhaust. Presented at: eighth Health Effects Institute annual conference; April; Colorado Springs,
17 CO. Cambridge, MA: Health Effects Institute.
18
19 Mauderly, J. L. (1994) Noncancer pulmonary effects of chronic inhalation exposure of animals to solid particles.
20 In: Mohr, U.; Dungworth, D. L.; Mauderly, J. L.; Oberdorster, G., eds. Toxic and carcinogenic effects
21 of solid particles in the respiratory tract: [proceedings of the 4th international inhalation symposium];
22 March 1993; Hannover, Germany. Washington, DC: International Life Sciences Institute Press;
23 pp. 43-55.
24
25 McLemore, T. L.; Martin, R. R.; Wray, N. P.; Cantrell, E. T.; Busbee, D. L. (1981) Reassessment of the
26 relationship between aryl hydrocarbon hydroxylase and lung cancer. Cancer (Philadelphia)
27 48: 1438-1443.
28
29 Medinsky, M. A.; Shelton, H.; Bond, J. A.; McClellan, R. O. (1985) Biliary excretion and enterohepatic
30 circulation of 1-nitropyrene metabolites in Fischer-344 rats. Biochem. Pharmacol. 34: 2325-2330.
31
32 Medinsky, M. A.; Bond, J. A.; Hunsberger, S.; Sun, J. D. (1988) Lung, liver, and kidney as potential target
33 organs after exposure to 1-nitropyrene, as determined by the time course of covalently bound material.
34 J. Toxicol. Environ. Health 23: 445-454.
35
36 Medinsky, M. A.; Bond, J. A.; Hunsberger, S.; Griffith, W. C., Jr. (1989) A physiologically based model of
37 1-nitropyrene metabolism after inhalation or ingestion. Health Phys. 57 (suppl. 1): 149-155.
38
39 Miles, P. R.; Lee, P.; Trash, M. A.; Van Dyke, L. (1977) Chemiluminescence associated with phagocytosis of
40 foreign particles in rabbit alveolar macrophages. Life Sci. 20: 165-170.
41
42 Mitchell, C. E. (1982) Distribution and retention of benzo(a)pyrene in rats after inhalation. Toxicol. Lett.
43 11:35-42.
44
45 Mitchell, C. E. (1988) Formation of DNA adducts in mouse tissues after intratracheal instillation of
46 1-nitropyrene. Carcinogenesis (London) 9: 857-860.
47
48 Moller, L.; Tornquist, S.; Beije, B.; Rafter, J.; Toftgard, R.; Gustafsson, J.-A. (1987) Metabolism of the
49 carcinogenic air pollutant 2-nitrofluorene in the isolated perfused rat lung and liver. Carcinogenesis
50 (London) 8: 1847-1852.
51
52 Moller, L.; Corrie, M.; Midtvedt, T.; Rafter, J.; Gustafsson, J.-A. (1988) The role of the intestinal microflora
53 in the formation of mutagenic metabolites from the carcinogenic air pollutant 2-nitrofluorene.
54 Carcinogenesis (London) 9: 823-830.
December 1994 10-38 DRAFT-DO NOT QUOTE OR CITE
-------
1 Morrow, P. E. (1986) The setting of paniculate exposure levels for chronic inhalation toxicity studies. J. Am.
2 Coll. Toxicol. 5: 533-544.
3
4 Morse, M. A.; Carlson, G. P. (1985) Distribution and macromolecular binding of benzo[a|pyrene in SENCAR
5 and BALB/c mice following topical and oral administration. J. Toxicol. Environ. Health 16: 263-276.
6
7 Nesnow, S.; Triplett, L. L.; Slaga, T. J. (1984) Tumor initiating activities of 1-nitropyrene and its nitrated
8 products in SENCAR mice. Cancer Lett. (Shannon, Irel.) 23: 1-8.
9
10 Nikula, K. J.; Snipes, M. B.; Barr, E. B.; Mauderly, J. L. (1991) Histopathology and lung tumor responses in
11 rats exposed to diesel exhaust or carbon black. In: Annual report of the Inhalation Toxicology Research
12 Institute operated for the United States Department of Energy by the Lovelace Biomedical and
13 Environmental Research Institute: October 1, 1990 though September 30, 1991. Albuquerque, NM:
14 Inhalation Toxicology Research Institute, Lovelace biomedical and Environmental Research Institute;
15 pp. 87-129; report no. LMF-134.
16
17 Nikula, K. J.; Snipes, M. B.; Barr, E. B.; Griffith, W. C.; Henderson, R. F.; Mauderly, J. L. (1994) Influence
18 of particle-associated organic compounds on the carcinogenicity of diesel exhaust. In: Mohr, U.;
19 Dungworth, D. L.; Mauderly, J. L.; Oberdorster, G., eds. Toxic and carcinogenic effects of solid
20 particles in the respiratory tract: [proceedings of the 4th international inhalation symposium]; March
21 1993; Hannover, Germany. Washington, DC: International Life Sciences Institute Press; pp. 565-568.
22
23 Oberdorster, G.; Yu, C. P. (1991) The carcinogenic potential of inhaled diesel exhaust: a particle effect?
24 J. Aerosol Sci. 21(suppl. 1): S397-S401.
25
26 Otofuji, T.; Horikawa, K.; Maeda, T.; Sano, N.; Izumi, K.; Otsuka, H.; Tokiwa, H. (1987) Tumorigenicity test
27 of 1,3- and 1,8-dinitropyrene in BALB/c mice. JNCI J. Natl. Cancer Inst. 79: 185-188.
28
29 Perwak, J.; Byrne, M.; Coons, S.; Goyer, M.; Harris, J.; Cruse, P.; DeRosier, R.; Moss, K.; Wendt, S. (1982)
30 An exposure and risk assessment for benzo[a]pyrene and other polycyclic aromatic hydrocarbons.
31 Volume IV. Benzo[a]pyrene, acenaphthylene, benz[a]anthracene, chrysene, dibenz[a,h]anthracene,
32 benzo[b]fluoranthene, benzo[k]fluoranthene, benzo[g,h,i]perylene, indeno[l,2,3-c,d]pyrene. Washington,
33 DC: U.S. Environmental Protection Agency, Office of Water Regulations and Standards; EPA report no.
34 EPA-440/4-85-020. Available from: NTIS, Springfield, VA; PB85-222586.
36 Phillips, B. J.; James, T. E. B.; Anderson, D. (1984) Genetic damage in CHO cells exposed to enzymically
37 generated active oxygen species. Mutat. Res. 126: 265-271.
38
39 Pitts, J. N., Jr.; Lokensgard, D. M.; Harger, W.; Fisher, T. S.; Mejia, V.; Schuler, J. J.; Scorziell, G. M.;
40 Katzenstein, Y. A. (1982) Mutagens in diesel exhaust paniculate: identification and direct activities of
41 6-nitrobenzo[a]pyrene, 9-nitroanthracene, 1-nitropyrene and 5//-phenanthro[4,5-fco/]pyran-5-one. Mutat.
42 Res. 103: 241-249.
43
44 Plopper, C. G.; Hyde, D. M.; Weir, A. J. (1983) Centriacinar alterations in lungs of cats chronically exposed to
45 diesel exhaust. Lab. Invest. 49: 391-399.
46
47 Pott, F.; Stober, W. (1983) Carcinogenicity of airborne combustion products observed in subcutaneous tissue and
48 lungs of laboratory rodents. Environ. Health Perspect. 47: 293-303.
49
50 Randerath, K.; Randerath, E.; Agarwal, H. P.; Gupta, R. C.; Schurdak, M. E.; Reddy, M. V. (1985)
51 Postlabeling methods for carcinogen-DNA adduct analysis. Environ. Health Perspect. 62: 57-65.
53 Reddy, M. V.; Randerath, K. (1986) Nuclease PI-mediated enhancement of sensitivity of 32P-postlabeling test for
54 structurally diverse DNA adducts. Carcinogenesis (London) 7: 1543-1551.
December 1994 10_39 DRAFT-DO NOT QUOTE OR CITE
-------
1 Rister, J.; Baehner, R. L. (1977) Effect of hyperoxia on superoxide anion and hydrogen peroxide production of
2 polymorphonuclear leucocytes and alveolar macrophages. Br. J. Haematol. 36: 241-248.
3
4 Root, R. K.; Metcalf, J. A. (1977) H2O2 release from human granulocytes during phagocytosis; relationship to
5 superoxide anion formation and cellular catabolism of H2O2: studies with normal and cytochalasin
6 B-treated cells. J. Clin. Invest. 60: 1266-1279.
7
8 Roy, A. K.; El-Bayoumy, K.; Hecht, S. S. (1989) 32P-postlabeling analysis of 1-nitropyrene-DNA adducts in
9 female Sprague-Dawley rats. Carcinogenesis (London) 10: 195-198.
10
11 Saffiotti, U.; Cefis, F.; Kolb, L. H.; Shubik, P. (1965) Experimental studies of the conditions of exposure to
12 carcinogens for lung cancer induction. J. Air Pollut. Control Assoc. 15: 23-25.
13
14 Sagai, M.; Saito, H.; Ichinose, T.; Kodama, M.; Mori, Y. (1993) Biological effects of diesel exhaust
15 particles. I. In vitro production of superoxide and in vivo toxicity in mouse. Free Radical Biol.
16 Med. 14: 37-47.
17
18 Sandusky, C. B.; Cowden, M. W.; Schwartz, S. L. (1977) Effect of particle size on regurgitative exocytosis by
19 rabbit alveolar macrophages. In: Sanders, C. L.; Schneider, R. P.; Dagle, G. E.; Ragan, H. A., eds.
20 Pulmonary macrophage and epithelial cells: proceedings of the sixteenth annual Hanford biology
21 symposium; September 1976; Richland, WA. Washington, DC: Energy Research and Development
22 Administration; pp. 85-105. (ERDA symposium series: no. 43). Available from: NTIS, Springfield, VA;
23 CONF-760927.
24
25 Sato, S.; Ohgaki, H.; Takayama, S.; Ochiai, M.; Tahira, T.; Ishizaka, Y.; Nagao, M.; Sugimura, T. (1986)
26 Carcinogenicity of dinitropyrenes in rats and hamsters. In: Ishinishi, N.; Koizumi, A.; McClellan, R. O.;
27 Stober, W., eds. Carcinogenic and mutagenic effects of diesel engine exhaust: proceedings of the
28 international satellite symposium on lexicological effects of emissions from diesel engines; July; Tsukuba
29 Science City, Japan. Amsterdam, The Netherlands: Elsevier Science Publishers B. V.; pp. 271-277.
30 (Developments in toxicology and environmental science: v. 13).
31
32 Schuetzle, D.; Riley, T. L.; Prater, T. J.; Harvey, T. M.; Hunt, D. F. (1982) Analysis of nitrated polycyclic
33 aromatic hydrocarbons in diesel particulates. Anal. Chem. 54: 265-271.
34
35 Sims, P.; Grover, P. L.; Swaisland, A.; Pal, K.; Hewer, A. (1974) Metabolic activation of benzo(a)pyrene
36 proceeds by a diol-epoxide. Nature (London) 252: 326-328.
37
38 Slaga, T. J.; Viaje, A.; Berry, D. L.; Bracken, W.; Buty, S. G.; Scribner, J. D. (1976) Skin tumor initiating
39 ability of benzo(a)pyrene 4,5- 7,8- and 7,8-diol-9,10-epoxides and 7,8-diol. Cancer Lett. (Shannon, Irel.)
40 2: 115-121.
41
42 Sone, S. (1986) Role of alveolar macrophages in pulmonary neoplasias. Biochim. Biophys. Acta 823: 227-245.
43
44 Sone, S.; Brennan, L. M.; Creasia, D. A. (1983) In vivo and in vitro NO2 exposures enhance phagocytic and
45 tumoricidal activities of rat alveolar macrophages. J. Toxicol. Environ. Health 11: 151-163.
46
47 Stenback, F.; Rowland, J.; Sellakumar, A. (1976) Carcinogenicity of benzo(a)pyrene and dusts in the hamster
48 lung (instilled intratracheally with titanium oxide, aluminum oxide, carbon and ferric oxide). Oncology
49 33: 29-34.
50
51 Stowers, S. J.; Anderson, M. W. (1985) Formation and persistence of benzo(a)pyrene metabolite-DNA adducts.
52 Environ. Health Perspect. 62: 31-39.
53
December 1994 10-40 DRAFT-DO NOT QUOTE OR CITE
-------
1 Strom, K. A. (1984) Response of pulmonary cellular defenses to the inhalation of high concentrations of diesel
2 exhaust. J. Toxicol. Environ. Health 13: 919-944.
3
4 Sun, J. D.; Wolff, R. K.; Kanapilly, G. M. (1982) Deposition, retention, and biological fate of inhaled
5 benzo(a)pyrene adsorbed onto ultrafine particles and as a pure aerosol. Toxicol. Appl. Pharmacol.
6 65: 231-244.
7
8 Sun, J. D.; Wolff, R. K.; Kanapilly, G. M.; McClellan, R. O. (1984) Lung retention and metabolic fate of
9 inhaled benzo(a)pyrene associated with diesel exhaust particles. Toxicol. Appl. Pharmacol. 73: 48-59.
10
11 Sun, J. D.; Bond, J. A.; Dahl, A. R. (1988a) Biological disposition of vehicular airborne emissions:
12 particle-associated organic constituents. In: Watson, A. Y.; Bates, R. R.; Kennedy, D., eds. Air
13 pollution, the automobile, and public health. Washington, DC: National Academy Press; pp. 299-322.
14
15 Sun, J. D.; Wolff, R. K.; Maio, S. M.; Barr, E. B. (1988b) The influence of adsorption to carbon black
16 particles on the retention and metabolic activation of benso(a)pyrene in rat lungs following inhalation
17 exposure or intratracheal instillation. Inhalation Toxicol. 1: 79-94.
18
19 Tatsumi, K.; Amano, H. (1987) Biotransformation of 1-nitropyrene and 2-nitrofluorene to novel metabolites, the
20 corresponding formylamino compounds, in animal bodies. Biochem. Biophys. Res. Commun.
21 142: 376-382.
22
23 Thyssen, J.; Althoff, J.; Kimmerle, G.; Mohr, U. (1981) Inhalation studies with benzo[a]pyrene in Syrian golden
24 hamsters. JNCI J. Natl. Cancer Inst. 66: 575-577.
25
26 Tyrer, H. W.; Cantrell, E. T.; Horres, R.; Lee, I. P.; Peirano, W. B.; Danner, R. M. (1981) Benzo(a)pyrene
27 metabolism in mice exposed to diesel exhaust: I. Uptake and distribution. Environ. Int. 5: 307-311.
28
29 Urban, J. L.; Shepard, H. M.; Rothstein, J. L.; Sugarman, B. J.; Schreiber, H. (1986) Tumor necrosis factor:
30 a potent effector molecule for tumor cell killing by activated macrophages. Proc. Natl. Acad. Sci.
31 U.S. A. 83: 5233-5237.
32
33 Vostal, J. J. (1986) Factors limiting the evidence for chemical carcinogenicity of diesel emissions in long-term
34 inhalation experiments. In: Ishinishi, N.; Koizumi, A.; McClellan, R. O.; Stober, W., eds. Carcinogenic
35 and mutagenic effects of diesel engine exhaust: proceedings of the international satellite symposium on
36 lexicological effects of emissions from diesel engines; July; Tsukuba Science City, Japan. Amsterdam,
37 The Netherlands: Elsevier Science Publishers B. V.; pp. 381-396. (Developments in toxicology and
38 environmental science: v. 13).
39
40 Warheit, D. B.; George, G.; Hill, L. H.; Snyderman, R.; Brody, A. R. (1985) Inhaled asbestos activates a
41 complement-dependent chemoattractant for macrophages. Lab. Invest. 52: 505-514.
42
43 Weinstein, I. B.; Troll, W. (1977) National Cancer Institute workshop on tumor promotion and cofactors in
44 carcinogenesis. Cancer Res. 37: 3461-3463.
45
46 Weiss, S. J.; Rustagei, P. K.; LoBuglio, A. F. (1978) Human granulocyte generation of hydroxyl radical.
47 J. Exp. Med. 147: 316-323.
48
49 Weitzman, S. A.; Stossel, T. P. (1981) Mutation caused by human phagocytosis. Science (Washington, DC)
50 212: 546-547.
51
52 White, H. J.; Garg, B. D. (1981) Early pulmonary response of the rat lung to inhalation of high concentration of
53 diesel particles. J. Appl. Toxicol. 1: 104-110.
54
December 1994 10-4i DRAFT-DO NOT QUOTE OR CITE
-------
1 Williams, A. O.; Flanders, K. C.; Saffiotti, U. (1993) Immunohistochemical localization of transforming
2 growth factor-/31 in rats with experimental silicosis, alveolar type II hyperplasia, and lung cancer.
3 Am. J. Pathol. 142: 1831-1840.
4
5 Wislocki, P. G.; Bagan, E. S.; Lu, A. Y. H.; Dooley, K. L.; Fu, P. P.; Han-Hsu, H.; Beland, F. A.;
6 Kadlubar, F. F. (1986) Tumorigenicity of nitrated derivatives of pyrene, benz[a]anthracene, chrysene,
7 and benzofajpyrene in the newborn mouse assay. Carcinogenesis (London) 7: 1317-1322.
8
9 Wolff, R. K.; Bond, J. A.; Sun, J. D.; Henderson, R. F.; Harkema, J. R.; Griffith, W. C.; Mauderly, J. L.;
10 McClellan, R. O. (1989) Effects of adsorption of benzo[a]pyrene onto carbon black particles on levels of
11 DNA adducts in lungs of rats exposed by inhalation. Toxicol. Appl. Pharmacol. 97: 289-299.
12
13 Wolff, R. K.; Bond, J. A.; Henderson, R. F.; Harkema, J. R.; Mauderly, J. L. (1990) Pulmonary inflammation
14 and DNA adducts in rats inhaling diesel exhaust or carbon black. Inhalation Toxicol. 2: 241-254.
15
16 Wong, D.; Mitchell, C. E.; Wolff, R. K.; Mauderly, J. L.; Jeffrey, A. M. (1986) Identification of DNA damage
17 as a result of exposure of rats to diesel engine exhaust. Carcinogenesis (London) 7: 1595-1597.
18
19 Wood, A. W.; Levin, W.; Lu, A. Y. H.; Yagi, H.; Hernandez, O.; Jerina, D. M.; Conney, A. H. (1976)
20 Metabolism of benzo[a]pyrene and benzo[a]pyrene derivatives to mutagenic products by highly purified
21 hepatic microsomal enzymes. J. Biol. Chem. 251: 4882-4890.
December 1994 10-42 DRAFT-DO NOT QUOTE OR CITE
-------
i 11. QUALITATIVE AND QUANTITATIVE
2 EVALUATION OF THE CARCINOGENICITY
3 OF DIESEL ENGINE EMISSIONS
4
5
6 11.1 INTRODUCTION
7 Concern about the carcinogenic risk of exposure to diesel engine emissions was
8 stimulated in the late 1970s by a report indicating that diesel exhaust is mutagenic (Huisingh
9 et al., 1978), by the knowledge that diesel exhaust contains known carcinogens, and by the
10 projected increase in the use of diesel engines in passenger vehicles. This concern
11 culminated in a U.S. Environmental Protection Agency (EPA)-sponsored quantitative cancer
12 risk estimate for diesel engine emissions (Albert et al., 1983). This study, however, did not
13 include either a qualitative evaluation of carcinogenicity or an evaluation of noncancer health
14 effects of diesel engine emissions. Their estimate, moreover, was based on a "comparative
15 potency" method because of a lack of either chronic animal cancer bioassays or definitive
16 epidemiological data.
17 Since 1983, several chronic animal inhalation studies and epidemiological investigations
18 designed to assess the carcinogenicity of diesel engine emissions have been completed.
19 These studies are summarized in Chapters 7 and 8. A variety of experiments carried out
20 with the goal of elucidating mechanisms of diesel exhaust-induced carcinogenicity have also
21 been published. Because of the increase in the availability of data and because of the need to
22 provide an up-to-date evaluation of the hazards of diesel exhaust inhalation for the Office of
23 Mobile Sources, a qualitative as well as a quantitative assessment of cancer risk was
24 undertaken. These assessments are presented in this chapter.
25 The qualitative evidence for carcinogenicity of diesel exhaust is evaluated in
26 Section 11.2. A summary of the International Agency for Research on Cancer (IARC)
27 conclusions is also included (International Agency for Research on Cancer, 1989). This is
28 followed by a review of previous quantitative risk estimates in Section 11.3. A discussion of
29 the exhaust constituents likely to be responsible for cancer induction and their influence on
30 cancer model selection is covered in Section 11.4. The approaches for quantitating risk are
31 outlined in Section 11.5; unit risk estimates are calculated, the results are discussed, and a
December 1994 !!_! DRAFT-DO NOT QUOTE OR CITE
-------
1 unit risk estimate recommended. Both qualitative and quantitative assessments are
2 summarized in Section 11.6.
3
4
5 11.2 WEIGHT OF EVIDENCE FOR CARCINOGENICITY OF DIESEL
6 EXHAUST
7 Lung cancer incidence has been studied in human populations exposed to diesel engine
8 exhaust. An increased incidence of lung cancer was observed in four mortality studies
9 (Howe et al., 1983; Bofetta and Stellman, 1988; Garshick et al., 1988; Wong et al., 1985)
10 and in seven case-control studies (Williams et al., 1977; Hall and Wynder, 1984; Damber
11 and Larson, 1987; Garshick et al., 1987; Benhamou et al., 1988; Hayes et al., 1989;
12 Steenland et al., 1990). A dose response trend was observed in three of the cohort studies
13 (Howe et al., 1983; Bofetta and Stellman, 1988; Wong et al., 1985) and two of the case-
14 control studies (Garshick et al., 1987; Steenland et al., 1990). An increased incidence of
15 lung cancer was not observed in some of the other studies (Waller, 1981; Rushton et al.,
16 1983; Wong et al., 1985; Edling et al., 1987; Lerchen et al., 1987), but each had several
17 methodological limitations such as small sample sizes, short follow-up, lack of adjustment for
18 confounding factors, etc. The studies reporting increased incidences also had some major
19 limitations even though more recent ones, especially those reported by Garshick and
20 co-workers, were able to eliminate most of the confounding variables.
21 Some epidemiological studies suggest an association between diesel exhaust exposure
22 and lung cancer. Because of the uncertainties created by limited exposure data and the
23 possibility of exposure to other agents, the evidence for carcinogenicity of diesel engine
24 emissions in humans is considered to be limited under Environmental Protection Agency's
25 cancer assessment guidelines (Federal Register, 1986).
26 In animal experiments, inhalation of whole diesel exhaust resulted in the induction of
27 lung tumors in F344 rats (Brightwell et al., 1986; Ishinishi et al., 1986; Iwai et al., 1986;
28 Mauderly et al., 1987), in Wistar rats (Heinrich et al., 1986b), in Sencar mice (Pepelko and
29 Peirano, 1983), and in NMRI mice (Heinrich et al., 1986b; Stober, 1986). Lung tumors
30 were also induced by implantation of diesel exhaust condensate in Osborne-Mendel rats
31 (Grimmer et al., 1987). Skin painting of diesel particle extracts induced dermal tumors in
December 1994 H_2 DRAFT-DO NOT QUOTE OR CITE
-------
1 strain "A" mice (Kotin et al., 1955) and in Sencar mice following promotion with
2 tetradecanoylphorbol-13-acetate (TPA) (Nesnow et al., 1982). Extensive studies with
3 Salmonella mutagenesis assays have unequivocally demonstrated direct-acting mutagenic
4 activity in both particle extracts and gaseous fractions of diesel exhaust. Positive results have
5 also been reported for gene mutations and chromosome effects in mammalian cell systems
6 after exposure to diesel particle extracts.
7 Based on the induction of lung tumors via inhalation in at least two strains of rats and
8 two strains of mice and by implantation of diesel exhaust condensate, subcutaneous tumors
9 following injection of exhaust particle organic extracts, and skin tumors following dermal
10 application of exhaust particle organic extracts and supported by positive mutagenicity
11 results, the evidence for carcinogenicity of diesel exhaust in animals is considered to be
12 sufficient under EPA's Cancer Assessment Guidelines (Federal Register, 1986).
13 The International Agency for Research on Cancer (1989) also evaluated the evidence
14 for carcinogenicity of diesel exhaust and concluded that the evidence for carcinogenicity in
15 humans is limited. This conclusion was based primarily on cohort studies of railroad
16 workers in the United States (Garshick et al., 1988) and Canada (Howe et al., 1983) and two
17 case-control studies (Garshick et al., 1988; Howe et al., 1983) using individuals drawn from
18 the same group of workers. Three further studies of cohorts with less certain exposure to
19 diesel engine exhaust were also considered: two studies of London bus company employees
20 (Rushton et al., 1983; Raffle, 1957) and one of Swedish dock workers (Edling et al., 1987).
21 The IARC concluded that the evidence for carcinogenicity of whole engine exhaust in
22 experimental animals was adequate. Their conclusions were based on positive tumorigenic
23 effects in two different strains of rats in five of six experiments (Karagianes et al., 1981;
24 Iwai et al., 1986; Ishinishi et al., 1986; Heinrich et al., 1986a; Mauderly et al., 1987) and
25 on positive effects in mice in two studies (Pepelko and Peirano, 1983; Stober, 1986).
26 The IARC conclusions are thus in general agreement with those of the EPA. Both
27 concluded that human evidence is limited. The IARC considered the animal evidence to be
28 adequate for whole diesel exhaust as well as for extracts of exhaust particles but inadequate
29 for the gaseous phase. Although EPA has not specifically evaluated the gaseous phase and
30 did not consider whole exhaust and particle extracts separately, both agencies agree that
31 whole diesel engine exhaust is carcinogenic in animals.
December 1994 n_3 DRAFT-DO NOT QUOTE OR CITE
-------
1 On the basis of limited evidence for carcinogenicity in humans, diesel engine emissions
2 are considered to best fit into cancer weight-of-evidence Category Bl, according to EPA
3 cancer assessment guidelines. This classification is supported by sufficient evidence in
4 animals and positive results in mutagenicity studies and is consistent with the presence of
5 known carcinogens on diesel particles. Chemicals classified in Category Bl are considered
6 to be probable human carcinogens. The International Agency for Research on Cancer (1989)
7 also considers diesel exhaust to be probably carcinogenic in humans.
8
9
10 11.3 REVIEW OF PREVIOUS QUANTITATIVE RISK ESTIMATES
11 Early attempts to assess quantitatively the carcinogenicity of diesel engine emissions
12 were hindered by a lack of both positive epidemiologic studies and long-term animal studies.
13 One means of overcoming these obstacles was the use of the "comparative potency" method
14 (Albert et al., 1983). A second involved estimating risk based on equivocal epidemiological
15 evidence (Harris, 1983).
16 In the comparative potency method, a combustion product was selected that had a
17 previously determined cancer potency estimate based on epidemiologic data. The ratios of
18 the potency of this agent (e.g., coke oven emissions) to diesel particulate matter extract in a
19 variety of in vivo and in vitro tests are then multiplied by the epidemiology-based potency
20 estimate for coke oven emissions and averaged. If epidemiology-based estimates from more
21 than one pollutant are used, the derived potencies are generally averaged to obtain an overall
22 mean.
23 The comparative potency estimate of Albert et al. (1983) is probably best known.
24 Their results were obtained using epidemiology-based unit cancer risk estimates for coke
25 oven emissions, cigarette smoke condensate, and roofing tar. Samples of particulate matter
26 were collected from three LDD engines (a Nissan 220 C, an Oldsmobile 350, and a
27 Volkswagen turbocharged Rabbit), all run on a highway fuel economy test cycle, and a
28 heavy-duty engine (Caterpillar 3304) run under steady-state, low-load conditions. The
29 particulate matter was extracted with dichloromethane and tested in a variety of assays.
30 Dose-dependent increases in response were obtained for the four assays listed below:
31
32
December 1994 11-4 DRAFT-DO NOT QUOTE OR CITE
-------
1 (1) Ames Salmonella typhimurium (TA98) reverse mutation,
2
3 (2) gene mutation in L5178Y mouse lymphoma cells,
4
5 (3) Sencar mouse skin tumor initiation test, and
6
7 (4) viral enhancement of chemical transformation in Syrian hamster embryo cells.
8
9
10 Only the first three assays were used for development of comparative potency estimates
11 because of variability of responses in the enhancement of viral transformation assay. The
12 in vitro studies were carried out both in the presence and absence of metabolic activators.
13 The potency, defined as the slope of the dose-response curve, was measured for each sample
14 in each short-term assay.
15 The skin tumor initiation test was positive for all the engines tested except the
16 Caterpillar engine. Only the Nissan engine, however, gave strong dose-response data.
17 Because this was considered to be the most biologically relevant test, it was used to derive
18 potency estimates for the Nissan engine. An estimate for the Nissan engine was then derived
19 by multiplying the epidemiology-based potency estimates for each of the three agents (coke
20 oven emissions, roofing tar, and cigarette smoke condensate) by the ratios of their potencies
21 in the skin tumor initiation test to that of the Nissan diesel engine. According to this
22 method, three 95% upper-bound estimates of lifetime cancer risk per micrograms per cubic
23 meter of extractable organic matter were derived for the Nissan diesel, based on potency
24 comparisons with each of the three agents. These values are coke oven emissions,
25 2.6 x 104; roofing tar, 5.2 x 10"4; and cigarette smoke condensates, 5.4 x 10"4. The
26 average of the three is equal to 4.4 x 10"4.
27 The potency of the other diesel emission samples were not estimated directly because of
28 the weak response in the skin tumor initiation test. Instead, their potency relative to the
29 Nissan engine was estimated as the arithmetic mean of their potency relative to the Nissan in
30 the Salmonella assay in strain TA98, the sister chromatic exchange (SCE) assay in Chinese
31 hamster ovary (CHO) cells, and the mutation assay in mouse lymphoma cells. The estimated
32 lifetime cancer risk per micrograms per cubic meter for extracts from these engines are as
33 follows: Volkswagen, 1.3 x lO"4; Oldsmobile 1.2 x 10"4; and Caterpillar, 6.6 x 10'6.
December 1994 H_5 DRAFT-DO NOT QUOTE OR CITE
-------
1 To convert these values to a lifetime risk per micrograms per cubic meter of total
2 particulate matter, the results were multiplied by the fraction of extractable organic matter in
3 the particles. This conversion was based on the assumption that the carcinogenic effects
4 were caused solely by the organic fraction. These fractions were Nissan, 0.08; Volkswagen,
5 0.18; Oldsmobile, 0.17; and Caterpillar, 0.27. After this adjustment, the resulting estimated
6 potencies per micrograms per cubic meter of inhaled diesel particules varied from 3.5 x 10"5
7 for the Nissan to 1.8 x 10'6 for the Caterpillar.
8 Harris (1983) developed comparative potency estimates for the same four engines used
9 by Albert et al. (1983) but used only two epidemiology based potency estimates, those for
10 coke oven emissions and for roofing tar. He used preliminary data from three of the same
11 assays as did Albert et al. (1983), the Sencar mouse skin tumor initiation assay, enhancement
12 of viral transformation in Syrian hamster embryo cells, and the L5178 mouse lymphoma test.
13 The mouse lymphoma test was used both with and without metabolic activation, whereas the
14 Salmonella assay was not used.
15 The diesel cancer potency estimates by Harris were then derived by multiplying the
16 epidemiology based cancer potency estimates for both coke oven emissions and roofing tar by
17 the ratio of their potencies compared with diesel exhaust particles in each of the four
18 bioassays. For example, the epidemiology-based relative risk of exposure to 1 /zg/m3 of
19 coke oven emissions was estimated to equal 4.4 x lO^/ig/m3. In the skin tumor initiation
20 test, 2.1 papillomas per mouse were reported for the coke oven sample, compared with
21 0.53 for the Nissan engine extract. The benzene extractable fraction was assumed to equal
22 0.06 (slightly less than that in the Albert et al., 1983) studies. The diesel potency estimate
23 using this comparison is then equal to (0.53/2.1) x 0.06 x 4.4 x 10"%ig/m3, or
24 6.6 x 10"6//ig/m3 particulate matter. A total of eight comparisons were made for each
25 engine, four bioassays times two epidemiology-based potency estimates.
26 The Harris (1983) estimates are not comparable to those of Albert et al. (1983) without
27 adjustment. The unit risk estimates of Albert et al. (1983) are based on absolute risk during
28 lifetime exposure, whereas Harris reported his values in terms of relative risk per year of
29 exposure. To adjust this to lifetime risk for continuous exposure, it is necessary to multiply
30 Harris's values by a factor of 2.7 = 70 x 0.039, where 70 reflects the lifetime exposure
31 (70 years), and 0.039 is the lifetime lung cancer mortality rate in the U.S. population.
December 1994 11-6 DRAFT-DO NOT QUOTE OR CITE
-------
1 The range of potencies varied from 0.2 x 10'5 to 0.6 x 10'5 for the Nissan sample
2 0.1 x 10'5 to 2.4 x 10'5 for the Oldsmobile 350, 0.2 x 10'5 to 27.8 x 10'5 for the
3 Volkswagen Rabbit, and 0.1 x 10"5 to 2.5 x 10'5//ig/m3 paniculate matter for the
4 Caterpillar sample. Harris (1983) derived an overall mean relative risk value of
5 3.5 x 10"5//ig/m3 for the three light-duty engines with a 95% upper confidence limit of
6 2.52 x 10"4. Individual mean values for each engine were not reported. After multiplying
7 by 2.7 to convert to a unit risk, the upper bound estimate of potency of the paniculate matter
8 from the three light-duty engines was equal to 6.8 x lO^/^g/m3. McClellan (1986),
9 Cuddihy et al. (1981, 1984), and Cuddihy and McClellan (1983) reported a unit risk of about
10 7.0 x 10"5//ig/m3 using a comparative potency method similar to those reported in the
11 preceding paragraph. The data base was similar to that used by Albert et al. (1983) and
12 Harris (1983). This estimate agrees quite well those reported by Albert et al. (1983).
13 Although the Harris (1983) estimate is somewhat greater, it should be remembered that it
14 was based on preliminary data.
15 With the availability of chronic cancer bioassays, more recent assessments were based
16 on lung tumor induction in rats. Albert and Chen (1986) reported a risk estimate based upon
17 the chronic rat bioassay conducted by Mauderly et al. (1987). Using a multistage model and
18 assuming equivalent deposition efficiency in humans and rats, they derived a 95% upper
19 confidence limit of 1.6 x 10"5 for lifetime risk of exposure to 1 /*g/m3. Pott and Heinrich
20 (1987) used a linear extrapolation, including data reported by Brightwell et al. (1986),
21 Heinrich et al. (1986b) and Mauderly et al. (1987). They reported risk estimates of 6 x 10'5
22 to 12 x 10~5/^ig/m3. Most recently, Smith and Stayner (1990), using time-to-tumor models
23 based on the data of Mauderly et al. (1987), derived 95% upper confidence limits ranging
24 from 1.5 x 10'5 to 3 x lO^g/m3.
25 At least two attempts were made to estimate lung cancer risk based upon epidemiology
26 data. Harris (1983) also assessed the risk of exposure to diesel engine emissions using data
27 from the London Transport Worker Study reported by Waller (1981). Five groups of
28 employees from the London Transport Authority (LTA) were used. These included bus
29 garage engineers, bus drivers, bus conductors, engineers in central works, motormen, and
30 guards. The first group was considered to have received the highest exposure; the next two,
31 intermediate; and the last two, none. When cancer death rates for the high-exposure group
December 1994 H_7 DRAFT-DO NOT QUOTE OR CITE
-------
1 were compared with those of London males, there was no increase in the observed to
2 expected (O/E) ratios. The author, in fact, considered the results to be negative. Because
3 the low rate of lung cancer in all the LTA exposure groups, however, may have been the
4 result of a "healthy worker" effect, (Harris, 1983) compared the exposed groups with
5 internal controls. He merged the three exposed groups and compared them with the two
6 groups considered to be unexposed. An adjustment was made for the estimated greater
7 exposure levels of garage engineers compared with bus drivers and conductors. Using this
8 method, the relative risk of the exposed groups was greater than 1, but was statistically
9 significant only for garage engineers exposed from 1950 to 1960.
10 Harris (1983) identified a variety of uncertainties relative to potency assessment based
11 on this study. These included
12
13
14 • small unobserved differences in smoking incidences among groups, which could have
15 a significant effect on lung cancer rates;
16
17 • uncertainty about the magnitude of exposure in the exposed groups;
18
19 • uncertainty regarding the extent of change in exposure conditions over time;
20
21 • random effects arising from the stochastic nature of the cancer incidence; and
22
23 • uncertainty in the mathematical specification of the model.
24
25
26 Taking the uncertainties into account, he derived a maximum likelihood relative risk
27 estimate of 1.23 x 10"4 with a 95% upper confidence limit of 5 x lO^/xg/m3 paniculate
28 matter per year. These estimates are equal to 5 x 10"4 and 2 x 10"3, respectively, when
29 converted to an absolute risk for lifetime exposure to 1 /xg/m3 paniculate matter. It should
30 be noted that, because of the high degree of uncertainty, the 95% lower confidence limit
31 would predict no risk.
32 More recently, McClellan et al. (1989) reported risk estimates based on the Garshick
33 et al. (1987) study in which lung cancer in railroad workers was evaluated. Using a
34 proportional risk model, 0.016 excess cancer deaths were estimated to occur for each year of
35 exposure to diesel exhaust. Adjustments were made to convert to continuous exposure
36 (168 versus 40 h) for 70 years. Because exposure levels could not be defined exactly, two
December 1994 11-8 DRAFT-DO NOT QUOTE OR CITE
-------
1 sets of calculations were made, assuming inhaled particle concentrations of either 500 or
2 125 /-tg/m3. Using an upper 95% confidence limit, the number of excess cancer deaths per
3 year in the United States were estimated to range from 1,900 to 7,400. These values could
4 then be converted to a lifetime risk of exposure to 1 /ig/m3 diesel exhaust ranging from
5 0.6 X 10"3 to 2 x 10"3. Even using the lower 95% confidence limits, an excess of 100 to
6 400 deaths are predicted, unlike the Harris study in which no excess deaths are predicted.
7 Based upon the lower 95% confidence interval, estimated lifetime risk ranges from
8 2.7 x 10"4 to 10 X 10"4. The estimates discussed in this section are listed in Table 11-1.
9
10
TABLE 11-1. ESTIMATED LIFETIME RISK OF CANCER FROM
INHALATION OF 1 /tg/m3 DIESEL PARTICIPATE MATTER
Method Potency" Comments Reference
Comparative potency 3.5 x 10'5 Nissan engine Albert et al. (1983)
Comparative potency 2.6 x 10"5 Average of three engines Albert et al. (1983)
Comparative potency 7.0 x 10'5 Light duty engines Cuddihy et al. (1984)
Comparative potency 6.8x10"* Average of three engines Harris (1983)
Multistage model 1.6 x 10'5 Lung cancer ratsb Albert and Chen (1986)
Straight line extrapolation 6-12 x 10"5 Lung cancer ratsc Pott and Heinrich (1987)
Time-to-tumor model 2-3 x 10'5 Lung cancer ratsb Smith and Stayner (1990)
Logistic regression 8 x 10'5 Lung cancer ratsd McClellan et al. (1989)
Epidemiological analysis 2 x 10'3 London transport study Harris (1983)
Epidemiological analysis 0.6-2 x 10'3 Railroad workers McClellan et al. (1989)
Estimated upper 95 % confidence limit of lifetime risk of continuous exposure to 1 /tg/m3 diesel exhaust
paniculate matter.
bUsed data from studies by Mauderly et al. (1987).
cUsed data from studies by Brightwell et al. (1986), Heinrich et al. (1986b), and Mauderly et al. (1987).
dUsed data from studies by Brightwell et al. (1986), Ishinishi et al. (1986), Iwai et al. (1986), and Mauderly
et al. (1987).
1 Each of the three methods used to assess cancer risk have important limitations. In the
2 case of the epidemiology-based estimates, exposure measurements were generally quite
3 limited. It was usually impossible to eliminate all confounding factors. Finally, relative risk
December 1994 n_9 DRAFT-DO NOT QUOTE OR CITE
-------
1 ratios were low, making them more sensitive to the effects of any confounding factors that
2 may be present.
3 With respect to the comparative potency method, it was assumed that the mix of
4 chemicals responsible for effects of coke oven emissions, cigarette smoke condensate, and
5 roofing tar, have the same relative potency in short-term tests as they do for cancer
6 induction. There is little direct evidence to support this assumption. Of even greater
7 importance is the fact that the comparative potency method is based upon test results from
8 organic extracts of diesel exhaust paniculate matter. Recent evidence, as discussed later in
9 the chapter, however, indicates that the inorganic particle core is likely to be primarily
10 responsible for the tumor induction. If this is true, then the biological basis for this method
11 has limited relevance for assessing the cancer potency of inhaled particulate matter.
12 The animal bioassay-based estimates contain uncertainties inherent in most species
13 extrapolations (i.e., possible differences in metabolism, target organ sensitivity, etc.). Dose
14 extrapolation methods were also relatively crude, and none attempted to estimate
15 concentration at the lung epithelial surface. For example, in only one estimate (Smith and
16 Stayner, 1990) was any adjustment made for species differences in particle deposition
17 efficiency. None of the previous risk estimates accounted for high dose inhibition of particle
18 clearance. Finally, previous estimates were based upon whole diesel exhaust, rather than the
19 fraction most likely to be responsible for cancer induction. In the present exercise, attempts
20 were made to determine which fraction of exhaust is responsible for lung cancer induction
21 and to accurately model the human target organ concentration of this fraction.
22
23
24 11.4 APPROACHES TO QUANTITATION OF HUMAN RISK FROM
25 EXPOSURE TO DIESEL EXHAUST
26 The objective of this section is to assess quantitatively the potential lung cancer risk to
27 humans resulting from exposure to diesel exhaust emissions in ambient air. Ideally,
28 prediction of human risk due to exposure to an environmental pollutant should be made on
29 the basis of human experience. Although several human epidemiological studies on bus,
30 dock, mine, and railroad workers are available, the data from these studies are not adequate
31 for assessing the potential cancer risk to humans from diesel exhaust exposure because of the
December 1994 1MO DRAFT-DO NOT QUOTE OR CITE
-------
1 lack of reliable information on exposure conditions these workers experienced. Therefore,
2 the challenge to risk assessors is how to provide a "best" risk assessment for diesel exhausts,
3 using all the available information from both animals and humans. In contrast to sparse
4 human data, there is rich information on diesel-induced lung tumors in two strains of rats.
5 An approach to integrate this diverse information is to perform a quantitative risk assessment
6 based on information from animals that includes bioassays and relevant biological
7 mechanisms, and then evaluate the animal-based results against available human experience.
8 This approach is adopted in this report.
9 In an attempt to quantitatively estimate risk using humans, a detailed analysis of the
10 Garshick et al. (1988) study on railroad workers was carried out. Garshick et al. (1988)
11 analyzed information obtained from the Railroad Retirement Board (RRB) on 55,407 white
12 males who began railroad employment between 1939 and 1949, who were between the ages
13 of 40 and 64 in 1959, and who in 1959 worked at one of the 39 jobs selected to represent a
14 range of potential diesel exhaust (DE) exposure. Two analyses that indicated an effect of
15 exposure to DE on lung cancer risk in this cohort were reported: (1) a relative risk for lung
16 cancer of 1.45 (95% CI = 1.11, 1.89) was observed for DE-exposed workers who were
17 40 to 44 years of age in 1959 and who consequently had the longest potential exposure to DE
18 (relative risk was progressively lower among DE-exposed workers who were older in 1959
19 and who had potentially shorter exposures to DE), and (2) the relative risk of lung cancer
20 increased monotonically with increasing duration of work in 1959 or later in a job involving
21 diesel exhaust exposure (disregarding exposures in the current year and in the most recent
22 4 years); this risk was 1.72 (95% CI = 1.29, 2.23) in the group with the longest exposure
23 (15 to 17 years).
24 The EPA sponsored an effort to provide a quantitative estimate of lung cancer risk
25 (Crump et al., 1991) on the basis of the epidemiological study by Garshick et al. (1988).
26 Their report is included in this document as Appendix B. More than 50 analyses of the
27 relationship between exposure to diesel exhaust and lung cancer incidence were conducted.
28 However, as detailed in Appendix B, none of these analyses demonstrated a pattern that was
29 consistent with an association between DE exposure and lung cancer; in fact, many of them
30 showed a statistically significant negative association.
December 1994 !!_!! DRAFT-DO NOT QUOTE OR CITE
-------
1 It should be pointed out that the failure to find a positive association between the degree
2 of diesel exhaust exposure and lung cancer in the Clement analyses (Crump et al., 1991)
3 does not necessarily negate the positive finding made by Garshick et al. (1988). In fact, the
4 first finding of Garshick et al. (when exposure was classified by "yes" or "no") was
5 supported by these analyses. It is possible that the exposure levels estimated for workers
6 were uncertain enough to obscure a weak but positive association between DE exposure and
7 lung cancer.
8 Another reason that the data may not be adequate for quantitative risk assessment is that
9 the Crump analyses revealed that there may be an undercount of deaths after 1977. This
10 possibility appears to be supported by a recently released data tape provided by the RRB that
11 includes follow-up of the cohort for additional years, as well as an update of the follow-up
12 through 1980. It is reported that for 1980, about 25% of the deaths on the updated tape
13 were not on the earliest tape (personal communication between Crump and Garshick, 1991).
14 Follow-up of railroad workers' mortality in the Garshick et al. (1988) study was through
15 1980, which is only 22 years since completion of the conversion of U.S. railroads to diesel.
16 Because the time from first exposure until evidence of an increased risk of environmentally
17 induced lung cancer is often on the order of 20 years, the full impact of DE on lung cancer
18 in this cohort may not be captured by the current study. Therefore, it would be worthwhile
19 to conduct a new study of this cohort to take advantage of the several years of follow-up
20 available. If such a study is conducted, it is recommended that vital status be verified
21 independently of the RRB record.
22 A second possible approach for developing a quantitative risk assessment would utilize
23 the comparative potency method. This approach was not selected for several reasons. First
24 of all it is not based upon actual measurements of lung cancer. As mentioned previously, it
25 was assumed that the mix of chemicals present in diesel exhaust and the pollutants to which
26 diesel exhaust was compared, have the same potency in short-term tests as they do for cancer
27 induction. The evidence to support this assumption is inadequate. Finally, the short-term
28 measurements used organic extracts of diesel particles rather than particles themselves.
29 As discussed later, the organic fraction, although contributing to, is not likely to be the
30 primary cause of lung cancer.
December 1994 11-12 DRAFT-DO NOT QUOTE OR CITE
-------
1 A third approach and the one selected utilized chronic animal cancer bioassays. This
2 approach was selected because several chronic studies have become available, adequate dose
3 response curves were reported, and the animals were exposed to whole exhaust, unlike the
4 comparative potency tests. Before the derivation of unit risk estimates, however, several
5 issues needed to be addressed. These included (1) determination of the critical target site for
6 diesel exhaust, (2) determination of the fraction of exhaust responsible for tumor induction,
7 (3) selection or development of dosimetric methods for accurately extrapolating dose from
8 experimental animals chronically exposed to high concentrations of exhaust to humans
9 exposed at ambient concentrations, and (4) selection of the most suitable low-dose
10 extrapolation model.
11 The critical target organ was considered to be the lung. Although Iwai et al. (1986) did
12 report the induction of malignant lymphomas as well as lung tumors in rats following diesel
13 exhaust exposure, the lung was the only target site in other experimental studies with this
14 species. Potential carcinogenic agents present in diesel exhaust may be absorbed from the
15 lungs, enter the bloodstream, and be transported systemically. Data, however, are lacking to
16 evaluate this possibility. Particle adsorbed organics may also reach systemic targets via the
17 GI tract. Particles deposited in the conducting airways are cleared to the mouth quite rapidly
18 and swallowed. A considerable volume of particles are also likely to be ingested as a result
19 of grooming during whole animal exposures (Wolff et. al., 1982) resulting in the possible
20 uptake of carcinogens by the gastrointestinal tract. Because half-times for elution of organics
21 from the particles are considerably longer than passage through the gastrointestinal tract,
22 however, the fraction absorbed is expected to be small. In any case, there is little evidence
23 for systemic effects of diesel exhaust.
24 The site of action in the lungs is assumed to be the epithelial lining of the alveoli and
25 small airways. According to Mauderly et al. (1987) inflammation and tumors appear to arise
26 from this tissue. Although a connection between interstitial events and lung tumors has been
27 suggested for particles (i.e., fibrosis as a precondition for lung tumors [Kuschner, 1984]),
28 data is unavailable to support this view with respect to diesel exhaust induced tumors.
29 Accurate extrapolation of dose from experimental studies using animals exposed at
30 high concentrations of exhaust to humans exposed to ambient concentrations requires a
31 variety of adjustments. These include adjustments for species differences in deposition
December 1994 H_13 DRAFT-DO NOT QUOTE OR CITE
-------
1 efficiency and respiratory exchange rates. One of the more important factors and one seldom
2 adjusted for in previous risk estimates is the rate of particle clearance from the deep lung.
3 Normal clearance half-times from the alveolar region are several times longer in humans than
4 rats (Chan et al., 1981; Bohning et al., 1982). This may result in an underestimate of lung
5 burden when extrapolating to humans. On the other hand, the high exposure concentrations
6 used in some of the animal studies resulted in a greatly slowed, or even a complete cessation
7 of clearance (Griffis et al., 1983). In order to accurately extrapolate dose from experimental
8 studies to humans, a detailed dosimetry model developed by Yu et al. (1991) was used. This
9 model is listed in Appendix D. The model accounts for species differences in respiratory
10 exchange rates, deposition efficiency, normal particle clearance rates, transport of particles to
11 lung associated lymph nodes and lung surface area. It also accounts for inhibition of particle
12 clearance due to lung overload. In this model dose is estimated in terms of particle
13 concentration per unit of lung surface area.
14 Two different approaches were used to derive unit risk estimates. The first one utilized
15 a linearized low-dose extrapolation model (LMS) and is based on the assumption that the
16 insoluble carbon core of the diesel particle is primarily responsible for the carcinogenic
17 effects of diesel exhaust. A particle based assessment was considered to be reasonable for
18 two principal reasons. First of all, exposure to the vapor phase alone did not result in
19 detectable tumor induction in rats (Brightwell et al., 1986; Mauderly et al., 1987; Iwai et al.,
20 1986; Stober, 1986). Secondly, exposure to carbon black, which is similar in composition to
21 the carbon core of whole diesel exhaust but contains only negligible amounts of organics was
22 about as effective in lung cancer induction as whole diesel exhaust (Heinrich, 1990;
23 Mauderly et al., 1991, Nikula et al., 1994). This issue is considered in more detail in the
24 discussion.
25 The LMS model is adopted by the EPA as a default procedure to provide an upper
26 bound estimate of risk when data useful to incorporate plausible mechanisms are not
27 available. The model was selected because, although mechanisms of carcinogenesis have
28 been proposed (see discussion section), they remain largely unproven. The Office of Health
29 and Environmental Assessment has developed several stochastic models that can be used to
30 incorporate biological mechanisms if such data are available.
December 1994 1M4 DRAFT-DO NOT QUOTE OR CITE
-------
1 A second approach is based upon the assumption that even though the concentration of
2 carcinogenic compounds on the diesel particles is small, they nevertheless can act in concert
3 with the particles to induce carcinogenesis. An alternative low-dose extrapolation model was
4 developed, which allows for the possibility that organic materials may induce organ specific
5 adducts which contribute to the carcinogenic process. This model was developed because it
6 was recognized that various PAHs and nitroaromatics are present, even if at low
7 concentrations, and may contribute to cell initiation. This model is outlined in detail in
8 Appendix C.
9 Quantitative assessments are presented in Section 11.5. Both approaches selected
10 utilize the detailed dosimetry model previously mentioned to estimate concentration of
11 particulate matter at airway and alveolar surfaces. Risk is based upon the assumption that an
12 equivalent concentration of particles per unit of lung surface area results in equivalent risk in
13 humans and rats. In one low dose extrapolation model, the linearized multistage, dose is
14 based on the carbon particle minus adsorbed organics and on the carbon particle with the
15 organics present in an alternative model.
16
17
18 11.5 DOSE-RESPONSE CALCULATIONS BASED ON ANIMAL
19 BIOASSAY DATA
20 Calculation of unit risk estimates, is provided in this section. Unit risk is a
21 quantification of the carcinogenic potency for the compound. The unit risk estimate for an
22 air pollutant is defined as the 95 % upper bound of the increased lifetime cancer risk for an
23 individual continuously exposed for a lifetime to an air pollutant at a concentration of
24 1 /xg/m3 in ambient air. The results of the unit risk calculations are summarized in
25 Table 11-6.
26
27 11.5.1 Data Available for Risk Calculations
28 As reviewed in Chapter 7, five bioassay studies showed positive lung tumor responses
29 in rats (Brightwell et al., 1986; Ishinishi et al., 1986; Iwai et al., 1986; Stober, 1986;
30 Mauderly et al., 1987). Only three of them (Tables 11-2 through 11-4) were used for unit
December 1994 H_15 DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 11-2. INCIDENCE OF LUNG TUMORS IN FISCHER 344 RATS
(MALES AND FEMALES COMBINED) EXPOSED TO
HEAVY-DUTY ENGINE EXHAUST
Exposure
Concentration
(mg/m3)
0
0.35
3.50
7.08
Dose
Weekly
Exposure
(mg/m3 x h)a
0
12
122
248
Metric
Lung Particle Burden
(mg/cm2 lung surface)b
0
6.4 x 10'5
2.8 X ID'3
6.0 x 10-3
Lung Tumor Incidence
2/230
3/223
8/222
29/227
"Exposures were 7 h/day, 5 days/week.
bCalculated using mathematical models in Appendix D.
Source: Mauderly et al. (1987).
TABLE 11-3. INCIDENCE OF LUNG TUMORS IN
FISCHER 344 RATS (MALES AND FEMALES COMBINED)
EXPOSED TO HEAVY-DUTY ENGINE EXHAUST
Exposure
Concentration
(mg/m3)
0
0.46
0.96
1.84
3.72
Dose
Weekly
Exposure
(mg/m3 x h)a
0
44
92
177
357
Metric
Lung Particle Burden
(mg/cm2 lung surface)15
0
2.5 x 10-4
2.0 x 10'3
4.2 x 10'3
8.8 x 10'3
Lung Tumor Incidence
1/123
1/123
0/125
4/123
8/124
Exposures were 16 h/day, 6 days/week.
bCalculated using mathematical models in Appendix D.
Source: Ishinishi et al. (1986).
December 1994
11-16
DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 11-4. INCIDENCE OF LUNG TUMORS IN
FISCHER 344 RATS (MALES AND FEMALES COMBINED)
EXPOSED TO DIESEL ENGINE EXHAUST
Exposure
Concentration
(mg/m3)
0
0.7
2.2
6.6
Weekly
Exposure
(mg/m3 x h)a
0
56
176
528
Dose Metric
Lung Particle Burden
(mg/cm2 lung surface)b
0
3.5 x 10-*
4.2 x 10'3
1.3 x ID'2
Lung Tumor Incidence
4/250
1/112
14/112
55/111
"Exposures were of 16 h/day, 5 days/week.
bCalculated using mathematical models listed in Appendix D.
cThe number of animals sacrificed at 6 and 12 mo are excluded from the denominators.
Source: Brightwell et al. (1986).
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
risk calculations because each study selected was designed using multiple exposure groups
studies of animals exposed to carbon black, although not used in the risk calculations, did
influence the methodology. The time-to-event (i.e., death with or without tumors) data are
available for the Mauderly et al. (1987) study. These time-to-event data are used in all the
risk calculations based on the Mauderly et al. (1987) data.
11.5.2 Calculation of Unit Risks
The linearized multistage (LMS) model used in approach number one to calculate particle
based risk estimates has the mathematical form P = 1 - exp(-Z), where Z is either
Z = q0 + qjxd + ...-l- qmxdm, a polynomial of concentration d; or Z = (Q0 +
Qjxd + ...+ Qmxdm) xtk, a polynomial of concentration d multiplied by a time factor fi and
thus is more appropriate for risk calculations. The availability of preliminary data from
when time-to-event data are used. When time-to-event data are used, the lifetime risk is
calculated by actuarial life table approach using the survival probability of the NTP control
animals (Fischer 344 rats) provided in Portier et al. (1986). The range of extrapolation is
about three orders of magnitude in the present study. Denote P0 the lifetime cancer risk at
December 1994
11-17 DRAFT-DO NOT QUOTE OR CITE
-------
1 concentration 0. Because the extra risk (P - P0)/(l - P0) is dominated by the linear term
2 qjxd at low concentration, the 95% upper bound of Q] is used to represent unit risk when
3 d is expressed in micrograms per cubic meter.
4 When extrapolating risk from animals to humans, a dose metric that will induce the
5 same tumor incidence rate in animals and humans must be assumed (i.e., dose equivalence
6 assumption). The method used for calculating equivalent doses uses a mathematical model to
7 adjust for the dosimetric parameters determining lung burden of particulate matter in rats and
8 humans and to correct to a dose per unit lung surface area. In this method, the dose per unit
9 lung surface area is considered to be equally potent in the induction of lung tumor responses
10 in both animals and humans. The deposition-clearance-retention model used to estimate dose
11 per unit lung surface area is described in Appendix D. This model accounts for animal-
12 human differences in regional deposition efficiency; respiratory exchange rates; particle
13 clearance rates at low doses, as well as at doses resulting in impaired clearance; and lung
14 surface area. In these calculations, the mass fraction of the total particle adsorbed organics
15 is assumed to be 20%, with half the mass composed of slowly eluted organics (t1/2 =
16 30 days) and the other half composed of rapidly eluted organics (t1/2 = 1.3 h). The
17 remainder is considered to be inorganic carbon.
18 The effect of high exposure concentrations with an accompanying impairment of
19 clearance on lung burden of organics is shown in Figure 11-1 and on lung burden of
20 particulate matter in Figure 11-2. The data are plotted in terms of estimated lung burden per
21 unit of exposure. Little or no overload is seen at the lowest exposure concentration. As a
22 result the modeled lung burdens of both organics and carbon particulate matter reach an
23 asymptote after a few weeks of exposure. At higher concentrations particle clearance slows
24 and may even stop. In this case, lung particles burdens increase continually during exposure.
25 The organic constituents, on the other hand, elute fairly quickly from the particles and reach
26 a steady state even during continued exposure at high concentrations. Lung burdens of
27 organics are therefore less affected by overload inhibition of clearance.
28 Most of the tumor incidence data in the animal experiments were recorded at exposure
29 concentrations resulting in some degree of particle clearance inhibition. Unless adjustment
30 for clearance rates are made, lung particle burdens during low-dose extrapolation will be
December 1994 11-18 DRAFT-DO NOT QUOTE OR CITE
-------
0.05-
c
'§ 0.04H
c r-—~.
O""1 0.03~
0.02-
co
O) 0.01 -
7.08 mg/m3
3.46 mg/m3
0.35 mg/m3
Organics
o.oo-.
0 20 40 60 80 100 120 140
Weeks of Exposure
Figure 11-1. Calculated lung burden (organic matter) in rats exposed (7 h/day,
5 days/week) to three different concentrations of particulate matter.
1 overestimated. On the other hand, because lung burdens of the organic fraction are less
2 influenced by clearance rates, they will decline at a differing rate from particles during low-
3 dose extrapolation. The fraction of exhaust selected for development of cancer potency
4 estimates can therefore affect results.
5 Determination of dose for the carbon core becomes problematic because the lung
6 burdens over time differ drastically between low- and high-exposure concentrations. As can
7 be seen in Figure 11-2, a steady state is reached only for the low-exposure concentrations.
8 The difference in the behavior of lung burden over time between low- and high-exposure
9 groups suggests that the use of a lung burden at a fixed time point (e.g., at 1 year after
10 exposure began) as an estimate of dose may not be appropriate. In this report, the average
11 lung burden (in milligrams) is used as the target organ dose. This is calculated by dividing
December 1994 1M9 DRAFT-DO NOT QUOTE OR CITE
-------
.0
"S
0
O "
o£
. D)
U&
ID
D
OQ
5-
4-
co
i
ro
i
1-
0
Carbon
7.08 mg/m3
3.47 mg/m3
0.35 mg/m3
0
i
20
40 60 80 100
Weeks of Exposure
120
Figure 11-2. Calculated lung burden (carbon core) in rats exposed (7 h/day,
5 days/week) to three different concentrations of particulate matter.
1 the area under the curve (of lung burden over time) by the corresponding time period
2 (156 weeks) over which area under the curve is calculated.
3 A second approach to estimating risk was attempted because it was considered more
4 desirable to base risk upon a biologically based dose-response model. Although the data are
5 presently insufficient to replace the linearized multistage model, the implications of
6 hypothetical mechanisms of cancer induction proposed at a 1992 EPA workshop on particles
7 can nevertheless be investigated. The biological issues considered included the role of
8 particle adsorbed organics as well as that of a variety of mediators secreted by particle laden
December 1994
11-20
DRAFT-DO NOT QUOTE OR CITE
-------
1 macrophages upon the carcinogenic process. A stochastic model was therefore developed
2 with the following properties:
3
4 (1) It accounts for the possible effects of both the carbon particles and its associated
5 organics.
6
7 (2) It allows evaluation of the contribution of both carbon particles and organics to
8 tumor induction.
9
10 (3) It allows for changing parameters with increasing lung burden.
11
12 (4) It was assumed that cell proliferation and tumor induction are stochastic. For
13 instance, it is not appropriate to assume that all cells divide at the same rate.
14
15
16 This model, which is illustrated in detail in Appendix C, allows for initiating properties
17 of both the carbon and the organic fraction and for the proliferative effects of the carbon
18 fraction. Although the mechanisms remain to be proven, it is assumed that carcinogenic
19 agents present in the organic fraction act directly upon the target cells, primarily via
20 initiation. It is further assumed that the majority of the particles are ingested by
21 macrophages. Particle laden macrophages are then induced to secrete a variety of mediators
22 (i.e., reactive oxygen species, cytokines, etc.), which diffuse to the target cells inducing
23 initiation, proliferation, and conversion of initiated cells to malignant cells.
24 There is considerable uncertainty regarding particle effects at low doses. It has been
25 claimed that particle-induced initiation and/or proliferation does not occur at low doses, (i.e.,
26 secretion of mediators only occurs when macrophages are overloaded with particles) (Vostal,
27 1986). There is inadequate evidence, at present, to support or refute this claim. Moreover,
28 even if macrophage overload is required, because of uneven distribution of particles, some
29 macrophages may become overloaded even at low exposure concentrations. Because of this
30 uncertainty the alternative model, like the LMS model, does not depart from linearity at low
31 doses.
32 The tumorigenic results reported by Mauderly et al. (1987) were the only data used to
33 estimate model parameters. Mauderly's data are the most useful because they contain
34 information on natural mortality and serial sacrifice of animals, a valuable tool for estimating
35 tumor latency. The model parameters are based upon the development of malignant tumors,
December 1994 H-21 DRAFT-DO NOT QUOTE OR CITE
-------
1 rather than all tumors as was done in the first method. This was necessary in order to utilize
2 mortality data. Lung burdens were calculated with the aid of the same dosimetry model used
3 for the linearized multistage (LMS) derived estimates.
4
5 11.5.3 Results of Unit Risk Calculations
6 Unit risks, based upon the LMS approach, were calculated using tumor incidence data
7 from each of the three bioassays (Brightwell et al., 1986; Ishinishi et al., 1986; and time-to-
8 event data from Mauderly et al., 1987) and the corresponding equivalent doses (Tables 11-2
9 through 11-4). The resultant unit risk estimates are listed in Table 11-5. They range from
10 1.6 x 10~5 to 7.1 x 10~5//ig/m3 with a geometric mean of 3.4 x 10"5//zg/m3. The unit risk
11 is defined as the 95% upper confidence limit of the risk of lung cancer mortality from
12 lifetime exposure to 1 /ig/m3 of diesel exhaust paniculate matter.
13
14
TABLE 11-5. UNIT RISK ESTIMATES PER MICROGRAMS PER
CUBIC METER OF DIESEL EXHAUST
95% Upper Confidence Limits of the Cancer Risk
from Exposure to 1 /xg/m3 Diesel Paniculate Matter
Mauderly et al. (1987) 3.4 x 10'5
Ishinishi et al. (1986) 1.6 x 10'5
Brightwell et al. (1986) 7.1 x 10'5
Geometric mean of three studies 3.4 x 10"5
alf milligrams per lung weight, instead of milligrams per lung surface is used as equivalent dose, those risk
estimates are reduced by a factor of 4.
1 In these calculations, the relationship between air concentration (in micrograms per
2 cubic meter) and lung burden (in milligrams) in humans is used to determine the human lung
3 burden resulting from inhalation of 1 ^ig/m3 of diesel exhaust paniculate matter. The particle
4 burden in terms of mass/unit lung surface area is then multiplied by the slope factor derived
5 from the animal data. For instance, when rat data from Ishinishi et al. (Table 11-3) are
6 used, the carcinogenic slope for rats (i.e., the 95% upper confidence limit of the linear
7 coefficient in the multistage model) expressed in terms of equivalent dose (micrograms of
December 1994 11-22 DRAFT-DO NOT QUOTE OR CITE
-------
1 carbon paniculate matter per square centimeter) is 8.3 x 10"3//*g/cm2. According to the
2 dosimetry model, an air concentration of 1 /xg/m3 of paniculate matter corresponds to a mass
3 of 1,230 jug of carbon particles per human lung. Because the lung epithelial surface area,
4 including the alveolar region and conducting airways, is assumed to equal 627,000 cm2, the
5 unit risk of 1.6 x 10'5/jtg/m3 is derived by multiplying 8.3 x 10'3//ig/cm2 x
6 1,230 jig/627,000 cm2.
7 The unit risk estimate derived using the alternate model and based upon the malignant
8 tumor data from the Mauderly et al. (1987) study is equal to 1.65 x 10'5//ig/m3. This is
9 lower than the estimate of 3.4 x 10"5/^g/m3 derived from the same study, based on the LMS
10 approach (Table 11-5). Application of the LMS model using only malignant tumors from
11 the Mauderly et al. study, rather than all lung tumors, however, resulted in a unit risk
12 estimate of 1.74 x 10"5//ig/m3. Thus, the unit risk estimates for the two approaches are
13 identical (1.7 x 10~5//*g/m3 when rounded to two significant figures). In fact, 1.7 cannot be
14 considered to differ from 3.4 x 10"3//ig/m3 because of the range of uncertainty. The
15 estimated risks may differ somewhat with increasing doses, since the slopes are not identical
16 at all exposure levels. See Appendix C for details.
17
18 11.5.4 Discussion of Unit Risk Estimates
19 11.5.4.1 Basis for the Present Approach
20 In most of the earlier evaluations, species differences in dosimetric parameters, such as
21 deposition efficiency and particle clearance rates were not accounted for. Of particular
22 importance, failure to account for overload inhibition of clearance as well as species
23 difference in lung particle clearance can result in erroneous estimates of low-dose lung
24 burdens in humans and thereby risk. Despite the greater uncertainty in the earlier estimates,
25 the ones based upon chronic animal bioassays do not differ greatly from the present one
26 (Table 11-1). They also fall within the same range as those derived using the comparative
27 potency method.
28 The human-epidemiology-based cancer risk estimates were, in most cases, greater than
29 animal-based ones. Exposure measurements, however, were quite limited in most of these
30 studies. Moreover, because of the very small increases in relative risk ratios, any degree of
31 confounding for smoking or other reasons can result in considerable error in the unit risk
December 1994 11-23 DRAFT-DO NOT QUOTE OR CITE
-------
1 estimates, Finally, as discussed previously, after an exhaustive evaluation of the Garshick
2 et al. (1987, 1988) epidemiology data, the analysis included in Appendix B was unable to
3 obtain positive dose response curves. The uncertainty in the previously published risk
4 estimates is therefore considered to be quite high.
5 The risk estimates using both the LMS model as well as the alternative model are based
6 on the dose equivalence assumption that the same lung surface concentration (in milligrams
7 per square centimeter) of particulate matter will lead to the same effect with respect to lung
8 tumor induction in both the rat and human lung epithelial cells. This is considered to be a
9 more accurate estimate of target organ dose then the one derived using the older EPA
10 approach. The use of surface area also was considered to be superior to lung weight because
11 lung tumors arise from epithelial cells rather than from interstitial cells.
12 If diesel exhaust induced lung tumors in humans originate in conducting airways rather
13 than alveoli, it could be argued that a dose based on surface area of conducting airways,
14 rather than the lower respiratory tract (LRT) surface, is more appropriate. Because it is
15 believed that tumor induction is related in some way to the particle-laden macrophages
16 residing in alveolar regions, it is assumed that exhaust-induced lung tumors in humans are
17 more likely to arise from the alveolar region. Thus, surface area of the LRT, which to a
18 large extent consists of alveolar surface, is considered to be the most appropriate dosimetric
19 parameter.
20 The risk estimates are also based upon the assumption that sensitivity of target organ
21 cells among species is equal. This is a departure from past EPA policy to adjust dose
22 according to body weight to the two-thirds power. Adjustment was based on the assumption
23 that slower rates of metabolism with accompanying slower rates of detoxification and/or
24 repair mechanisms renders the human cell more sensitive than cells from a smaller mammal.
25 It is unlikely in the current situation, however, that the primary cause of cancer is via direct-
26 acting carcinogens. More likely, epithelial cell transformation is induced by various factors
27 secreted by the particle-laden macrophages, possibly in conjunction with partially activated
28 organic agents. In this case, the greater sensitivity of human lung epithelial cells may be
29 offset as a result of slower production of these transforming agents and/or chemical
30 activation by the human alveolar macrophage. Obviously, these assumptions have not been
31 proven and are subject to challenge. The actual agent or agents responsible for the
December 1994 11-24 DRAFT-DO NOT QUOTE OR CITE
-------
1 diesel-induced lung tumor induction and the mechanisms of action have still not been
2 determined with any degree of certainty. Nevertheless, the evidence to date does not support
3 a further correction based on metabolic rate. Results of future research may establish an
4 appropriate factor, which could then be included.
5 Neither approach accounts for the possibility of cancer induction by the vapor phase
6 chemicals. There is little evidence to support the likelihood that this component contributes
7 significantly to the tumorigenic effects. Although benzene (Schuetzle and Frazier, 1986) as
8 well as aldehydes including acetaldehyde, acrolein, benzaldehyde, and formaldehyde (Smith,
9 1989) are contained in this fraction, the concentration of these known or suspected
10 carcinogens are quite low and unlikely to induce detectable tumorigenic responses at the
11 exhaust dilutions used. The sites of tumor induction by these chemicals also differ from
12 those induced by diesel exhaust. Benzene induces leukemia in humans (Rinsky et al., 1981),
13 zymbal gland tumors in rats and mice, lymphomas in mice and oral tumors in rats (Huff
14 et al., 1989). Aldehydes generally induce nasal tumors (Swenberg et al., 1980). Lung
15 tumors, however, have not been reported in any of these studies. Finally, exposure to
16 exhaust filtered to remove particles generally did not result in a detectable increase in tumor
17 incidence in rats (Brightwell et al., 1986; Mauderly et al., 1987).
18 Although halogenated dioxins (PCDDs) and dibenzofurans (PCDFs) are emitted in
19 many combustion processes, output from diesel engine are uncertain. Marklund et al. (1990)
20 were unable to measure any PCDDs or PCDFs in diesel engine emissions. Their detection
21 limits, however, were quite high due to technical difficulties. Jones (1993) cited data from
22 the California Air Resources Board, in which a diesel bus and a heavy duty truck were
23 reported to emit 1.6 and 4.9 ng/km, respectively. Oehme et al. (1991) estimated that heavy
24 duty diesel powered vehicles emitted between 0.8 and 9.5 ng/km PCDD/PCDF based upon
25 measured air concentrations along with differential counts of diesel and gasoline powered
26 vehicles in a highway tunnel. Results of such an indirect method, however, must be
27 considered to be highly uncertain.
28 There is even less data on ambient levels of PCDDs and PCDFs due to engine
29 emissions. Concentrations of PCDD/PCDFs were reported to be less than one picogram per
30 cubic meter at the tunnel exit in the Oehme et al. (1991) study. Because concentrations
31 would be expected to be considerably lower along open roads, these agents are unlikely to be
December 1994 H_25 DRAFT-DO NOT QUOTE OR CITE
-------
1 present in sufficient quantities to contribute significantly to adverse health effects from diesel
2 exhaust.
3 Although the vapor phase components failed to induce lung cancer in rats, the
4 possibility of interactive effects remains. Heinrich et al. (1982), for example, reported data
5 suggesting that the gaseous fraction of diesel exhaust promoted the tumorigenic effects of
6 dimethylnitrosamine in Syrian hamsters. Unfortunately, exposure of animals to diesel
7 exhaust paniculate matter alone, via inhalation, have not been carried out due to technical
8 difficulties. Because there are insufficient data regarding possible interactions with the
9 paniculate matter fraction, it is not considered practical, at this time, to attempt to account
10 separately for possible effects of the gaseous fraction. It should be noted, however, that
11 because concentration of the gaseous component generally varies with the paniculate matter
12 phase, it is at least partially accounted for in the risk analysis.
13 Until recently, the particle adsorbed organics were considered to be the primary source
14 of carcinogenicity in diesel exhaust. An early study, Kotin et al. (1955), reported that the
15 organic compounds extracted from the surface of the diesel particles were capable of tumor
16 induction. As detailed in Chapter 9, there is clear evidence that the organic constituents have
17 the capacity to interact with DNA to give rise to mutations, chromosome alterations, and cell
18 transformations, all well-established steps in the process of carcinogenesis. Furthermore, as
19 noted in Chapter 2, the organic chemicals present include a variety of PAHs and
20 nitroaromatics, many of which are known to be carcinogenic. These organics are eluted
21 from the particles with a shorter half-time than clearance of the particles themselves from the
22 deep lung (Sun et al., 1984). Following elution from the particles, the organics may diffuse
23 into the alveolar spaces and be taken up by susceptible lung cells, although amounts might be
24 limited by absorption into the bloodstream, metabolism by macrophages, etc.
25 The DNA adduct studies conducted by Bond et al. (1989) and detailed in Chapter 10
26 provided some evidence of a role for the organic moieties in tumor induction. For example,
27 diesel exhaust exposure induced adduct formation at paniculate matter concentrations of
28 3.5 mg/m3. Carbon black having only traces of organics also induced increases in adducts,
29 but only at concentrations three times as high (Bond et al., 1989). It is possible that the
30 carbon black effect seen at the higher dose was macrophage mediated, whereas the adducts
31 induced at lower doses of diesel exhaust were induced by organic constituents. If both
December 1994 11-26 DRAFT-DO NOT QUOTE OR CITE
-------
1 fractions induce adducts, then it is more likely that both are involved in the carcinogenic
2 process.
3 Most of the adducts occurred in peripheral lung tissue where tumors are found. Bond
4 et al. (1989) also reported a 60% increase in lung adduct levels of rats exposed to a diesel
5 particle concentration of 8.1 mg/m3, compared with no significant increases in the more
6 tumor—resistant hamster and mouse lungs, lending further support to a link between adduct
7 formation and subsequent tumor formation. It remains to be proven, however, whether the
8 adducts induced are specific for the mutational changes responsible for induction of cancer.
9 Additional evidence suggesting a role for the organic phase was provided by Mumford
10 et al. (1989). They reported greatly increased lung cancer mortality rates in Chinese
11 communes burning so called "smoky coal" but not wood or smokeless coal, in unvented open
12 pit fires used for heating and cooking. Particle concentrations ranged from 10 to 25 mg/m3
13 in communes burning either smoky coal or wood, but PAH concentrations were five to six
14 times greater in the air of communes burning smoky coal. Thus, cancer mortality correlated
15 more closely with concentrations of PAHs than with particles. In the case of smokeless coal,
16 both particle and PAH concentration were low. Demonstration of the carcinogenicity of coke
17 oven emissions in humans (Lloyd, 1971) also provided evidence for a role by the organics,
18 because coke oven paniculate matter lacks an insoluble carbon core.
19 Unquestionably, the organic phase of diesel exhaust contains known carcinogenic
20 compounds, most of which are aromatics. As can be seen in Table 3-5, however, the
21 concentration of poly cyclic aromatic compounds (PAHs) are quite low in diesel exhaust.
22 For example, all of the PAHs measured by Tong and Karasek (1984) account for only 25 to
23 50 /ig/mg of extract. If it is assumed that 20% of the paniculate matter mass is composed of
24 organic matter, then the total concentration of PAHs ranges from 5 to 10 /ig/mg of
25 paniculate matter. Furthermore, many of the PAHs are not known to be carcinogenic. The
26 concentration of the best known carcinogenic PAH (B[a]P) is present at concentrations of no
27 more than 0.1 /ng/mg paniculate matter, based upon the Tong and Karasek (1984) data. It is
28 unlikely that such low concentrations could be responsible for the degree of tumorigenic
29 responses seen at the concentrations of diesel exposures employed. Further support of a
30 minor role for organics is provided by the report that pyrolyzed pitch condensate, having
31 about three orders of magnitude greater concentration of organics than diesel exhaust
December 1994 H_27 DRAFT-DO NOT QUOTE OR CITE
-------
1 particles, but no insoluble particle core, was only about as potent as diesel exhaust in the
2 induction of cancer (Heinrich et al., 1986a).
3 A hypothesis first proposed by Vostal (1986) is based on the belief that "diesel"
4 particles themselves induce lung cancer, probably through a secondary effect. In support of
5 this hypothesis are the observations that lung cancer can be induced by inhalation of so-called
6 "inert" particles such as titanium dioxide (Lee et al., 1986) and coal dust (Martin et al.,
7 1977) or by intratracheal instillation of activated carbon (Kawabata et al., 1986). Heinrich
8 (1990) reported that inhalation of carbon black particles, which are virtually devoid of
9 organics and are similar to the carbon core of diesel exhaust, also induced lung cancer in rats
10 and did so at concentrations comparable to those used in the diesel exhaust studies.
11 Mauderly et al. (1991, 1993) reported similar results. Finally, Kawabata et al. (1994)
12 induced lung tumors in rats intratracheally instilled with diesel exhaust particulate matter
13 from which the organic components had been extracted. The extracted particles were
14 effective at doses comparable to those for unextracted particles. A third possibility is that
15 both particles and organics are important for tumor induction because the interaction of the
16 two fractions may result in an enhanced effectiveness of the organics. Stenback et al.
17 (1976), for example, reported that intratracheal instillation of B[a]P adsorbed to particles is
18 much more effective than in its pure form.
19 The interaction of particles with organics could occur in several ways, although
20 evidence is still lacking. Adsorption to particles may result in more effective penetration of
21 the organics into alveolar regions, where clearance is slower. If the organics are condensed
22 onto particles, they are more likely to be taken up by macrophages, where possible partial
23 activation of the carcinogenic PAHs may occur. In fact, Bond et al. (1984) provided
24 evidence that alveolar macrophages from beagle dogs metabolized B[a]P coated on diesel
25 exhaust particles to proximate carcinogenic forms. The relatively slow sustained release of
26 organics from the particles may also provide a relatively constant supply of xenobiotics to the
27 target cells.
28 The alternative model differs from the LMS approach in that both the particle as well
29 as the particle adsorbed organics are assumed to have a role in the carcinogenesis process.
30 The usefulness of the alternative model is based upon its ability to compare estimated risks
31 after varying the coefficients for initiation and/or proliferation of organic and particulate
December 1994 11-28 DRAFT-DO NOT QUOTE OR CITE
-------
1 matter phases. For example, if the coefficient for initiation by the organic phase is reduced
2 to zero, the risk is only reduced modestly (78% compared to the original model). On the
3 other hand, if the coefficient for proliferation by paniculate matter is increased to only
4 1.4 times that in the original model, the estimated risk nearly doubles. See Appendix C for
5 details. Although cell proliferation appears to have a greater influence upon cancer induction
6 in this model, this likelihood remains unproven since the coefficients for each parameter in
7 the model cannot be accurately determined at this time. If data becomes available to
8 accurately determine the proper coefficients, then it will be possible to account for the
9 relative contribution of both the particle and vapor phases for cancer induction.
10 Until the coefficients in the alternative model are determined, or other means of
11 partitioning risk among the components of diesel exhaust are developed, a unit risk based
12 upon paniculate matter and use of the linearized multistage model is considered to be the
13 most reasonable. This conclusion was reached not only because of the uncertainty in
14 determining coefficients in the alternate model, because of the demonstrated ability of pure
15 carbon to induce lung tumors and because of the low concentration of known carcinogens in
16 the organic.
17 Another important issue concerns extrapolation of tumor responses to ambient
18 concentrations. As can be seen from Tables 11-2 through 11-4, significant increases in lung
19 tumor incidences were not detected at the low exposure concentrations used in each of these
20 studies. Based upon exposure rates, the data might be interpreted as evidence for a threshold
21 in tumorigenic response. When the tumor responses were plotted in terms of modeled lung
22 burden (Figure 11-3), however, a change in slope at the low doses was not apparent.
23 In fact, for each of these studies, the data showed a very good statistical fit to a straight line.
24 The effect of clearance rates can be more clearly seen in Figure 11-4, where modeled lung
25 burdens in the three studies used for risk estimation are plotted against exposure rates. The
26 lung burden/exposure rate ratios are much lower at doses not inducing clearance inhibition.
27 Because the lung burdens were so small at the low exposure rates, detectable tumorigenic
28 responses would not be predicted.
29 Although the relationship between lung particle burden and tumor response admittedly
30 does not provide proof for the absence of a threshold, or a change in slope of the dose-
31 response curve at low exposures, it also fails to provide evidence to the contrary.
December 1994 H_29 DRAFT-DO NOT QUOTE OR CITE
-------
6 9 12
Lung Particle Burden (ug/cm2)
15
Figure 11-3. Relationship between lung tumor incidence and modeled lung particle
burden/unit of lung surface area using data from Brightwell et al. (1986),
Ishinishi et al. (1986), and Mauderly et al. (1987).
1 Nevertheless, since it appears that the primary means of cancer induction is macrophage
2 mediated, the possibility of nonlinear responses at low concentrations exists. Unfortunately,
3 information regarding mechanisms of action are inadequate. Based upon available data, lung
4 injury appears likely to be induced by release of mediators from activated macrophages
5 (e.g., reactive oxygen species, chemotactic factors, lysosomal hydrolases, other proteinases,
6 prostaglandins, plasminogen activators, and growth factors. Oberdorster and Yu (1991) have
7 outlined a hypothetical biological model, incorporating these factors, to explain the toxic and
8 carcinogenic effects of particles.
9 The relationship between macrophage particle burden and function may be quite
10 complex. The secretion of various mediators by individual macrophages may be proportional
11 to particle load, may be disproportional with no threshold, or may have a threshold. The
12 particles, moreover, are unevenly distributed among macrophages. Thus, some macrophages
13 may be overloaded even at low exposure levels. Furthermore, it is uncertain if the target
14 organ response to the mediators is linear or nonlinear at low doses. Because changes in the
December 1994
11-30
DRAFT-DO NOT QUOTE OR CITE
-------
120
240
360
480
600
Exposure Rate (mg • m3 • h/week)
Figure 11-4. Relationship between exposure rate and lung particle burden/unit of lung
surface area using data from Brightwell et al. (1986), Ishinishi et al.
(1986), and Mauderly et al. (1987).
1 dose-response curve in the cancer bioassays cannot be detected when dose is converted to
2 lung burden, and because data regarding release as well as effects of the supposed factors
3 mediating cancer induction are inadequate, it is considered inappropriate, at this time, to
4 attempt to adopt other than a linearized low-dose extrapolation model.
5 The LMS approach is also considered to be prudent and not likely to be overly
6 conservative for a number of other reasons as well. Even though the organics may not have
7 a major role in tumor induction, they cannot be totally discounted. Furthermore, any
8 initiating effects of organics are likely to be a straight line function of dose even at very low
9 doses. Nondiesel paniculate matter is present in ambient air at widely varying
10 concentrations. Although it cannot be assumed that all particles have similar cancer inducing
11 properties, nevertheless ambient paniculate matter does contribute to macrophage particle
December 1994
11-31
DRAFT-DO NOT QUOTE OR CITE
-------
1 burdens. There is also evidence that particle clearance in humans is slowed in smokers
2 (Bohning et al., 1982) and in individuals with respiratory disease (Cohen et al., 1979;
3 Freedman et al., 1988). As a result, there may be a large population already at the
4 threshold, or above, for lung particle overload. Because the lung burden model assumes
5 normal clearance at ambient particle concentrations, individuals with impaired clearance may
6 be near the upper bound estimates of risk. Bond et al. (1989) reported that levels of DNA
7 adducts were greater in the lungs of monkeys than in the lungs of rats exposed to the same
8 concentration of diesel exhaust, suggesting the possibility of increased sensitivity in a species
9 genetically more similar to humans. Finally, higher, although admittedly highly uncertain
10 risk estimates, derived from epidemiology data suggest that a conservative approach would
11 be prudent when extrapolating risk from animal data.
12 Attempts have been made to determine what characteristics of insoluble particles are
13 responsible their effects. As can be seen in Table 11-6, the concentrations of particles
14 required to induce tumors varied widely with particle type. Oberdorster and Yu (1991)
15 plotted tumor response against various parameters of the retained particles (i.e., volume,
16 weight, particle numbers, particle surface area, etc.). They determined that only the surface
17 area of the retained particles correlated well with observed lung tumors in the rat studies.
18 Assuming a particle surface area of 90 m2/g for diesel particles after the elution of organics
19 (Pierson and Brachaczek, 1976), the diesel cancer data fit a surface area-tumor incidence plot
20 quite well.
21 Although it is reasonable to assume that particle surface plays an important role in the
22 toxicity of inhaled insoluble paniculate matter, other particle characteristics cannot be ruled
23 out. In a recent study by Sagai et al. (1993), for example, the toxicity of diesel particulate
24 matter was claimed to be also related to the presence of oxygen radicals on the particle
25 surface. In any case, the cancer potency estimates for diesel exhaust should not be used to
26 estimate risk from exposure to other types of insoluble particulate matter found in the
27 ambient environment, because of their likely differing characteristics. In the present
28 assessment it is assumed that experimental animals and humans are exposed to the same types
29 of particles, so particle characteristics such as surface area do not need to be adjusted for.
December 1994 11-32 DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 11-6. CANCER STUDIES WITH RATS EXPOSED TO RELATIVELY
CHEMICALLY INERT DUSTS AT EXPOSURE CONCENTRATIONS OF SEVERAL
MILLIGRAMS PER CUBIC METER OR ABOVE
Reference
Martin et al. (1977)
Lee et al. (1986)
Heinrich (1993)
Muhle et al. (1991)
Heinrich (1990)
Mauderly et al. (1991)
Particle
Type
Coal dust
TiO2
TiO2a
Toner particles
Carbon black
Carbon black
Concentration
(mg/m3)
200
10-250
11.3
1-16
6
6.5
Duration
Not reported
2 years
2 years
2 years
10-20 mo
24 mo
Cancer
Yes
Yes (250 mg/m3)
Yes
No
Yes
Yes
aUltrafme particles.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
11.5.4.2 Evaluation of Animal-Based Risk Estimates Against Human Experience
It is of interest to evaluate the reasonableness of the bioassay-based risk estimates
(which range from 1.6 to 7.1 x 10"5 with a geometric mean of 3.4 x 10"5) against human
experience. Three sets of data offer such an opportunity: the Harris (1983) analysis of an
epidemiological study conducted on London Tranport Authority workers by Waller (1981)
and Garshick et al. (1987, 1988) on U.S. railroad workers. Although the human data are
considered to be unsuitable as a basis for calculating a unit risk (mainly due to the lack of
reliable exposure information), they can be used to evaluate the reasonableness of the animal-
based unit risk estimates.
Several attempts were made to estimate potential cancer risk due to exposure to diesel
exhaust on the basis of epidemiological data. Based on the London transport study, Harris
(1983) estimated that the increase in relative risk of lung cancer associated with 1 /ig/mVyear
of diesel exhaust exposure is 1.2 x 10"4 with a 95% upper bound of 4.8 x 10"4. As a
general practice when using data from a nonpositive epidemiological study to estimate cancer
risk, the unit risk is based upon the 95% upper bound. The resultant unit risk estimate was
equal to 2 x 10'3 (4.8 x 10"4 x 70 ^g/m3-years x 0.06), which is about 60-fold higher
than the mean animal-based unit risk estimate of 3.4 x 10"5 or about 30-fold greater than the
upper end of the range of animal-based unit risk estimates, 7.1 x 10"5. The animal-based
December 1994
11-33
DRAFT-DO NOT QUOTE OR CITE
-------
1 risk estimate is not considered to be inconsistent with this human-based estimate because
2 2 x 10~3 is an upper bound estimate from a nonpositive epidemiological study.
3 More recently, McClellan et al. (1989) reported risk estimates based on the Garshick
4 et al. (1987) study in which lung cancer in railroad workers was evaluated. Assuming
5 exposure concentrations of 500 and 125 jtg/m3, 95% upper bounds for lifetime cancer risk
6 per micrograms per cubic meter were estimated to be 6 x 10"4 and 2 X 10"3, respectively.
7 The smaller value of these two unit risk estimates is only about an order of magnitude
8 greater than the animal-based ones.
9 An epidemiological study that is potentially more suitable for quantitative risk
10 assessment was reported by Garshick et al. (1988). This study included a large number of
11 subjects and a small but significant increase in lung cancer mortality risk for some
12 subcohorts. Recently, the EPA has supported an effort to derive a unit risk estimate using
13 Garshick et al. (1988) and exposure data estimated by Woskie et al. (1988a,b; see Appendix
14 B for details). The data were analyzed in a variety of different ways by using relative risk
15 and absolute risk dose-response models and by classifying individuals into various exposure
16 categories according to job classification, duration of employment, age, different exposure
17 markers, etc. Even though at least 50 different analyses were carried out, an adequate dose-
18 response relationship could not be obtained.
19 The lack of a dose-response relationship is not totally unexpected, given the smallness
20 of mortality rate increase observed in the Garshick study, and great uncertainty associated
21 with the exposure estimates. The exposure data used in these studies came from air samples
22 collected during a limited time period (between 1981 and 1983) at four small railroads
23 operating in a limited geographical area (northern United States). Measured concentrations
24 were of respirable particulate matter rather than diesel exhaust per se, and these measured
25 concentrations were adjusted to produce markers of diesel exposure. These data were used
26 to estimate exposures to diesel exhausts that occurred among railroad workers throughout the
27 United States as much as 30 or more years ago. Diesel equipment and working conditions
28 have changed since the 1940s when the use of large numbers of diesel engines first began.
29 (Woskie et al. (1988b) described anecdotal reports of smoky working conditions in the diesel
30 repair shops during the 1950s and 1960s. They also reported that limited data available on
December 1994 11-34 DRAFT-DO NOT QUOTE OR CITE
-------
1 nitrogen oxide levels during these periods provide qualitative support for the likelihood of
2 higher levels of diesel exhaust concentration in the early years of dieselization. By the time
3 samples were collected for this study, these smoky conditions would have been largely
4 mitigated through improved ventilation and the advent of less smoky engines.
5 Based on Garshick's study, the relative risks of lung cancer death in exposed versus
6 unexposed railroad workers (classified into five subcohorts by ages in 1959) ranged from
7 0.96 with 95% confidence interval (0.74, 1.3) to 1.49 with 95% confidence interval (1.1,
8 1.9) The highest relative risk of 1.49 was observed from the workers who were 40 to
9 49 years old in 1959. This highest relative risk was used to evalute the reasonableness of the
10 animal-based unit risk estimate. It was assumed that this subcohort of workers was exposed
11 to diesel exhaust 8 h/day, 5 days/week, from age 35 to 65. The background lung cancer
12 mortality rate for this subcohort is estimated to be 0.038 (0.63 x 0.06), based on the fact
13 that the unexposed workers in the same age group had a relative risk of 0.63, and that the
14 corresponding lifetime lung cancer mortality rate in the general U.S. white male population
15 is about 0.06. If the lung cancer risk due to 1 /ig/m3 is assumed to be 3.4 x 10"5, it would
16 imply that the diesel concentration in their work environment was at least 400 /ig/m3. This
17 is calculated as follows: The risk due to 1 /ig/cm2 of particle lung burden is 0.017 (which
18 results in a unit risk of 3.4 x 10'5). To have a lower bound relative risk of 1.1, one would
19 need a lung burden (d) ug/cm2 that satisfies the relationship 0.1 x 0.038 = 0.017d (i.e.,
20 d = 0.22 /^g/cm2, which is equivalent to an air concentration of 0.4 mg/m3 by using the
21 dosimetry model of Yu et al., 1991). If the highest unit risk estimate 7.1 x 10"5 derived
22 from the Brightwell et al. (1986) study is used, then it would require at least 0.2 mg/m3 of
23 diesel exhaust concentration to observe a statistically significant elevation of lung cancer
24 mortality rate in this study. These air concentrations appear reasonable in light of working
25 conditions described by Woskie et al. (1988a).
26 Although the animal-based risk estimates are lower than those derived from human
27 data, they are not inconsistent for the following reasons:
28
29 (1) The human-base risk numbers are not derived from all available data; the numbers
30 are based only on a subset of data that showed the highest response. Therefore, the
31 resultant risk estimates are expected to be higher than if the whole data set is used.
32
December 1994 H_35 DRAFT-DO NOT QUOTE OR CITE
-------
1 (2) When a single data point (i.e., an overall relative risk and an averaged exposure
2 concentration) is used in the calculations, the resultant potency slope always results
3 in an overestimate of the slope factor if the dose-response relationship is not a
4 straight line over all exposure concentrations. To see this, one can assume (as an
5 example) that the cancer response follows a simple multistage model
6 P(d) = 1 - exp[-(q0 -I- qjd + q2d2)]. The relative risk at a concentration "d" is
7 R(d) = P(d)/P(0). Using the mathematical expression of R(d), it is easy to
8 demonstrate that the slope factor calculated by [R(d) - l]P(0)/d is always greater at
9 higher doses (which include the averged concentration used in the risk calculation)
10 than at low doses where the dose-response function is dominated by qt. A similar
11 (but not identical) concern of using averaged data has long been recognized by
12 epidemiologists who are concerned about the fact that ecologic or group-level
13 associations are not necessarily consistent with those measured at the individual
14 level (see Greenland and Robins, 1994a,b; Piantadosi, 1994; Cohen, 1994).
15
16 (3) There is evidence that various occupational groups were exposed to considerably
17 higher concentrations of diesel exhaust in the past than presently. For example,
18 particle concentration in a Finnish roundhouse was reported to average 2 mg/m2
19 (Heino et. al., 1978). This will result in greater lung burdens than predicted from
20 present exposures and, thereby, greater unit risk estimates.
21
22 (4) It is a reasonable assumption that the occupationally exposed groups on which the
23 unit risks were based were also exposed to greater nondiesel dust concentrations in
24 their earlier employment. This would again result in increased lung burdens and
25 possibly contribute to an overestimate of cancer risk.
26
27 11.5.4.3 Reasonableness of the Unit Risk Estimate
28 The dose-response analysis contained in this assessment of cancer risk has the following
29 positive features.
30
31 (1) The estimates are based upon several well designed, long-term animal studies.
32
33 (2) Epidemiology studies indicate that humans are susceptible to tumor induction by
34 inhalation of diesel exhaust.
35
36 (3) Dosimetry modeling, especially the portion accounting for high-dose inhibition of
37 particle clearance, has allowed more accurate extrapolation of dose from animals to
38 humans.
39
40 (4) Dose is based upon actual concentration of paniculate matter per unit lung surface
41 area.
42
43 (5) Use of an alternative model that attempted to account for initiation by the organic
44 fraction did not result in an appreciable change in the unit risk estimate.
December 1994 11-36 DRAFT-DO NOT QUOTE OR CITE
-------
1 Nevertheless, a number of uncertainties remain, the more significant ones of which are
2 listed below:
3
4
5 (1) In any species extrapolation there is the possibility of an inherent difference in
6 sensitivity to the agent being assessed.
7
8 (2) The selection of the inorganic paniculate matter fraction to base risk on. Although
9 evidence strongly supports the likelihood that particles are the major cause of lung
10 cancer induction by diesel exhaust, the organic constituents probably do contribute
11 at least minimally to the carcinogenic response. Because low-dose extrapolation of
12 organic matter lung burden is less influenced by lung overload than particles, the
13 unit risk estimate may differ somewhat if risk is base upon the organic fraction.
14 Nevertheless, a unit risk estimate based upon the organic fraction using the same
15 data sets, LMS and dosimetry model would differ by less than 25% from the
16 particle based one. Furthermore, use of the alternative model, incorporating
17 possible organic effects, as well as particle effects also failed to provide a
18 significantly different unit risk estimate.
19
20 (3) A departure from most EPA cancer risk assessments (Federal Register, 1986) is the
21 assumption of equivalent sensitivity across species based upon concentration per
22 unit of lung surface area. Although it is reasonable to assume that slower
23 metabolism by human lung epithelial cells is offset by slower production and
24 release of harmful factors, this assumption is still unproven.
25
26 (4) Use of linearized low-dose extrapolation methods. It is still uncertain whether
27 macrophages secrete mediators thought to induce cancer in lung epithelial cells at
28 particles burdens less than those necessary to induce inhibition of clearance. Even
29 if macrophages are activated at low particle burdens, it is uncertain if responses of
30 epithelial cells are linear at very low concentration. This area is considered to be a
31 primary research need.
32
33 (5) Heavy smokers and individuals exposed occupationally to dusty environments may
34 have lung particle burdens at or near threshold levels for inhibition of particle
35 clearance. The dose-response curve in such cases may be steeper than predicted by
36 extrapolation modeling, which is based upon normal clearance rates.
37
38
39 Overall, the unit risk estimate is considered to be reasonable because it is based upon
40 well-designed and conducted chronic inhalation experiments, human doses are estimated
41 using detailed modeling techniques, and results agree within an order of magnitude with unit
42 risks developed using a variety of approaches. Nevertheless, there is still a considerable
43 possibility that risk may be overestimated for some populations and underestimated in others.
44
December 1994 11.37 DRAFT-DO NOT QUOTE OR CITE
-------
1 11.6 SUMMARY AND CONCLUSIONS
2 As a result of limited evidence from epidemiological data, supported by adequate
3 evidence for carcinogenicity of diesel engine emissions in animal studies, as well as positive
4 evidence for mutagenicity, it was concluded that diesel engine emissions best fit into cancer
5 weight-of-evidence Category Bl. Diesel engine emissions are thus considered to be probable
6 human carcinogens. This is in agreement with a 2A classification by International Agency
7 for Research on Cancer.
8 Using a dosimetry model that accounted for animal-to-human differences in lung
9 deposition efficiency, lung particle clearance rates, lung surface area, ventilation, metabolic
10 rate, as well as elution rates of organic chemicals from the particle surface, equivalent human
11 doses were calculated on the basis of particle concentration per unit lung surface area.
12 Following dosimetric adjustment, risk estimates were derived using a linearized multistage
13 model. A unit risk estimate of 3.4 x 10~5 (the upper 95% bound of the cancer risk from
14 lifetime exposure to 1 /xg/m3 diesel particulate matter) is recommended. This estimate is
15 based on the geometric mean of estimates derived from three separate animal bioassays using
16 Fischer 344 rats.
17 This unit risk estimate should not be used to evaluate the cancer risk of other types of
18 particulate matter present in the ambient air. These particles may have differing solubilities,
19 surface areas, presence of free radicals, or other properties which may greatly affect cancer
20 potency.
December 1994 11-38 DRAFT-DO NOT QUOTE OR CITE
-------
1 REFERENCES
2
3 Albert, R. E.; Chen, C. (1986) U.S. EPA diesel studies on inhalation hazards. In: Ishinishi, N.;
4 Koizumi, A.; McClellan, R. O.; Stober, W., eds. Carcinogenicity and mutagenicity of diesel
5 engine exhaust: proceedings of the international satellite symposium on toxicological effects of
6 emissions from diesel engines; July; Tsukuba Science City, Japan. Amsterdam, The Netherlands:
7 Elsevier Science Publishers B. V.; pp. 411-419. (Developments in toxicology and environmental
8 science: v. 13).
9
10 Albert, R. E.; Lewtas, J.; Nesnow, S.; Thorslund, T. W.; Anderson, E. (1983) Comparative potency method for
11 cancer risk assessment: application to diesel paniculate emissions. Risk Anal. 3: 101-117.
12
13 Benhamou, S.; Benhamou, E.; Flamant, R. (1988) Occupational risk factors of lung cancer in a French
14 case-control study. Br. J. Ind. Med. 45: 231-233.
15
16 Boffetta, P.; Stellman, S. D. (1988) Association between diesel exhaust exposure and multiple myeloma:
17 an example of confounding. Prev. Med. 17: 236-237.
18
19 Bohning, D. E.; Atkins, H. L.; Cohn, S. H. (1982) Long-term particle clearance in man: normal and impaired.
20 In: Walton, W. H., ed. Inhaled particles V: proceedings of an international symposium; September 1980;
21 Cardiff, Wales. Ann. Occup. Hyg. 26: 259-271.
22
23 Bond, J. A.; Butler, M. M.; Medinsky, M. A.; Muggenburg, B. A.; McClellan, R. O. (1984) Dog pulmonary
24 macrophage metabolism of free and particle-associated [14C]benzo[a]pyrene. J. Toxicol. Environ. Health
25 14: 181-189.
26
27 Bond, J. A.; Harkema, J. R.; Henderson, R. F.; Mauderly, J. L.; McClellan, R. O.; Wolff, R. K. (1989)
28 Molecular dosimetry of inhaled diesel exhaust. In: Mohr, U.; Bates, D. V.; Dungworth, D. L.; Lee,
29 P. N.; McClellan, R. O.; Roe, F. J. C., eds. Assessment of inhalation hazards: integration and
30 extrapolation using diverse data: [papers from the 2nd international inhalation symposium]; 1989;
31 Hannover, FRG. Berlin, Federal Republic of Germany: Springer-Verlag; pp. 315-324.
32
33 Brightwell, J.; Fouillet, X.; Cassano-Zoppi, A.-L.; Gatz, R.; Duchosal, F. (1986) Neoplastic and functional
34 changes in rodents after chronic inhalation of engine exhaust emissions. In: Ishinishi, N.; Koizumi, A.;
35 McClellan, R. O.; Stober, W., eds. Carcinogenic and mutagenic effects of diesel engine exhaust:
36 proceedings of the international satellite symposium on toxicological effects of emissions from diesel
37 engines; July; Tsukuba Science City, Japan. Amsterdam, Holland: Elsevier Science Publishers B. V.;
38 pp. 471-485. (Developments in toxicology and environmental science: v. 13).
39
40 Chan, T. L.; Lee, P. S.; Hering, W. E. (1981) Deposition and clearance of inhaled diesel exhaust particles in the
41 respiratory tract of Fischer rats. J. Appl. Toxicol. 1: 77-82.
42
43 Cohen, B. L. (1994) Invited commentary: in defense of ecologic studies for testing a linear-no threshold theory.
44 Am. J. Epidemiol. 139: 765-768.
45
46 Cohen, D.; Arai, S. F.; Brain, J. D. (1979) Smoking impairs long-term dust clearance from the lung. Science
47 (Washington, DC) 204: 514-517.
48
49 Crump, K. S.; Lambert, T.; Chen, C. (1991) Assessment of risk from exposure to diesel engine emissions.
50 Research Triangle Park, NC: U.S. Environmental Protection Agency; contract no. 68-02-4601.
52 Cuddihy, R. G.; McClellan, R. O. (1983) Evaluating lung cancer risks from exposures to diesel engine exhaust.
53 Risk Anal. 3: 119-124.
54
December 1994 H_39 DRAFT-DO NOT QUOTE OR CITE
-------
1 Cuddihy, R. G.; Griffith, W. C.; Clark, C. R.; McClellan, R. 0. (1981) Potential health and environmental
2 effects of light duty diesel vehicles II. Albuquerque, NM: Lovelace Biomedical and Environmental
3 Research Institute, Inhalation Toxicology Research Institute; report no. LMF-89.
4
5 Cuddihy, R. G.; Griffith, W. C.; McClellan, R. O. (1984) Health risks from light-duty diesel vehicles. Environ.
6 Sci. Technol. 18: 14A-21A.
7
8 Damber, L. A.; Larsson, L. G. (1987) Occupation and male lung cancer: a case-control study in northern
9 Sweden. Br. J. Ind. Med. 44: 446-453.
10
11 Edling, C.; Anjou, C.-G.; Axelson, O.; Kling, H. (1987) Mortality among personnel exposed to diesel exhaust.
12 Int. Arch. Occup. Environ. Health 59: 559-565.
13
14 Federal Register. (1986) Guidelines for carcinogen risk assessment. F. R. (September 24) 51: 33992-34003.
15
16 Freedman, A. P.; Robinson, S. E.; Street, M. R. (1988) Magnetopneumographic study of human alveolar
17 clearance in health and disease. Ann. Occup. Hyg. 32(suppl. 1): 809-820.
18
19 Garshick, E.; Schenker, M. B.; Munoz, A.; Segal, M.; Smith, T. J.; Woskie, S. R.; Hammond, S. K.; Speizer,
20 F. E. (1987) A case-control study of lung cancer and diesel exhaust exposure in railroad workers.
21 Am. Rev. Respir. Dis. 135: 1242-1248.
22
23 Garshick, E.; Schenker, M. B.; Munoz, A.; Segal, M.; Smith, T. J.; Woskie, S. R.; Hammond, S. K.; Speizer,
24 F. E. (1988) A retrospective cohort study of lung cancer and diesel exhaust exposure in railroad workers.
25 Am. Rev. Respir. Dis. 137: 820-825.
26
27 Greenland, S.; Robins, J. (1994a) Invited commentary: ecologic studies—biases, misconceptions, and
28 counterexamples. Am. J. Epidemiol. 139: 747-760.
29
30 Greenland, S.; Robins, J. (1994b) Accepting the limits of ecologic studies: Drs. Greenland and Robins reply to
31 Drs. Piantadosi and Cohen. Am. J. Epidemiol. 139: 769-771.
32
33 Griffis, L. C.; Wolff, R. K.; Henderson, R. F.; Griffith, W. C.; Mokler, B. V.; McClellan, R. O. (1983)
34 Clearance of diesel soot particles from rat lung after a subchronic diesel exhaust exposure. Fundam.
35 Appl. Toxicol. 3: 99-103.
36
37 Grimmer, G.; Brune, H.; Deutsch-Wenzel, R.; Dettbarn, G.; Jacob, J.; Naujack, K.-W.; Mohr, U.; Ernst, H.
38 (1987) Contribution of polycyclic aromatic hydrocarbons and nitro-derivatives to the carcinogenic impact
39 of diesel engine exhaust condensate evaluated by implantation into the lungs of rats. Cancer Lett.
40 (Shannon, Irel.) 37: 173-180.
41
42 Hall, N. E. L.; Wynder, E. L. (1984) Diesel exhaust exposure and lung cancer: a case-control study. Environ.
43 Res. 34: 77-86.
44
45 Harris, J. E. (1983) Diesel emissions and lung cancer. Risk Anal. 3: 83-100.
46
47 Hayes, R. B.; Thomas, T.; Silverman, D. T.; Vineis, P.; Blot, W. J.; Mason, T. J.; Pickle, L. W.; Correa, P.;
48 Fontham, E. T. H.; Schoenberg, J. B. (1989) Lung cancer in motor exhaust-related occupations.
49 Am. J. Ind. Med. 16: 685-695.
50
51 Heino, M.; Ketola, R.; Makela, P.; Makinen, R.; Niemala, R.; Starck, J.; Partanen, T. (1978) Work conditions
52 and health of locomotive engineers: noise, vibration, thermal climate, diesel exhaust constituents,
53 ergonomics. Scand. J. Work Environ. Health 4: 3-14.
54
December 1994 11-40 DRAFT-DO NOT QUOTE OR CITE
-------
1 Heinrich, U. (1990) Results of long-term inhalation exposure of rats to carbon black "Printex 90" [letter to
2 Dr. Lester D. Grant). Presented at: U.S. Environmental Protection Agency peer review workshop on the
3 Health Assessment Document for Diesel Emissions; July; Research Triangle Park, NC.
4
5 Heinrich, U. (1993) Carcinogenic effects of solid particles. In: Toxicity and carcinogenicity of solid particles in
6 the respiratory tract—proceedings of the 4th international inhalation symposium; Hannover, pp. 43.
7
8 Heinrich, U.; Peters, L.; Funcke, W.; Pott, F.; Mohr, U.; Stober, W. (1982) Investigation of toxic and
9 carcinogenic effects of diesel exhaust in long-term inhalation exposure of rodents. In: Lewtas, J.,
10 ed. lexicological effects of emissions from diesel engines: proceedings of the Environmental Protection
11 Agency diesel emissions symposium; October 1981; Raleigh, NC. New York, NY: Elsevier Biomedical;
12 pp. 225-242. (Developments in toxicology and environmental science: v. 10).
13
14 Heinrich, U.; Pott, F.; Rittinghausen, S. (1986a) Comparison of chronic inhalation effects in rodents after
15 long-term exposure to either coal oven flue gas mixed with pyrolized pitch or diesel engine exhaust.
16 In: Ishinishi, N.; Koizumi, A.; McClellan, R. O.; Stober, W., eds. Carcinogenic and mutagenic effects
17 of diesel engine exhaust: proceedings of the international satellite syposium on lexicological effects of
18 emissions from diesel engines; July; Tsukuba Science City, Japan. Amsterdam, Holland: Elsevier Science
19 Publishers B. V.; pp. 441-457. (Developments in toxicology and environmental science: v. 13).
20
21 Heinrich, U.; Muhle, H.; Takenaka, S.; Ernst, H.; Fuhst, R.; Mohr, U.; Pott, F.; Stober, W. (1986b) Chronic
22 effects on the respiratory tract of hamsters, mice, and rats after long-term inhalation of high
23 concentrations of filtered and unfiltered diesel engine emissions. J. Appl. Toxicol. 6: 383-395.
24
25 Howe, G. R.; Fraser, D.; Lindsay, J.; Presnal, B.; Yu, S. Z. (1983) Cancer mortality (1965-77) in relation to
26 diesel fume and coal exposure in a cohort of retired railway workers. JNCI J. Natl. Cancer Inst.
27 70: 1015-1019.
28
29 Huff, J. E.; Haseman, J. K.; DeMarini, D. M.; Eustis, S.; Maronpot, R. R.; Peters, A. C.; Persing, R. L.;
30 Chrisp, C. E.; Jacobs, A. C. (1989) Multiple site carcinogenicity of benzene in Fischer 344 rats and
31 B6C3F; mice. Environ. Health Perspect. 82: 125-163.
32
33 Huisingh, J.; Bradow, R.; Jungers, R.; Claxton, L.; Zweidinger, R.; Tejada, S.; Bumgarner, J.; Duffield, F.;
34 Waters, M.; Simmon, V. F.; Hare, C.; Rodriguez, C.; Snow, L. (1978) Application of bioassay to the
35 characterization of diesel particle emissions. In: Waters, M. D.; Nesnow, S.; Huisingh, J. L.; Sandhu,
36 S. S.; Claxton, L., eds. Application of short-term bioassays in the fractionation and analysis of complex
37 environmental mixtures: [proceedings of a symposium; February; Williamsburg, VA]. New York, NY:
38 Plenum Press; pp. 383-418. (Hollaender, A.; Probstein, F.; Welch, B. L., eds. Environmental science
39 research: v. 15).
40
41 International Agency for Research on Cancer. (1989) Diesel and gasoline engine exhausts and some nitroarenes.
42 Lyon, France: World Health Organization; pp. 41-185. (IARC monographs on the evaluation of
43 carcinogenic risks to humans: v. 46).
44
45 Ishinishi, N.; Kuwabara, N.; Nagase, S.; Suzuki, T.; Ishiwata, S.; Kohno, T. (1986) Long-term inhalation
46 studies on effects of exhaust from heavy and light duty diesel engines on F344 rats. In: Ishinishi, N.;
47 Koizumi, A.; McClellan, R. O.; Stober, W., eds. Carcinogenic and mutagenic effects of diesel engine
48 exhaust: proceedings of the international satellite symposium on lexicological effects of emissions from
49 diesel engines; July; Tsukuba Science City, Japan. Amsterdam, Holland: Elsevier Science Publishers
50 B. V.; pp. 329-348. (Development in lexicology and environment science: v. 13).
51
52
December 1994 H_41 DRAFT-DO NOT QUOTE OR CITE
-------
1 Ishinishi, N.; Inamasu, T.; Hisanaga, A.; Tanaka, A.; Hirata, M.; Ohyama, S. (1988) Intratracheal instillation
2 study of diesel paniculate extracts in hamsters. In: Diesel exhaust and health risks: results of the HERP
3 studies. Tsukuba, Ibaraki, Japan: Japan Automobile Research Institute, Inc., Research Committee for
4 HERP Studies; pp. 209-216.
5
6 Ishinishi, N.; Kuwabara, N.; Takaki, Y.; Nagase, S.; Suzuki, T.; Nakajima, T.; Maejima, K.; Kato, A.;
7 Nakamura, M. (1988) Long-term inhalation experiments on diesel exhaust. In: Diesel exhaust and health
8 risks: results of the HERP studies. Tsukuba, Ibaraki, Japan: Japan Automobile Research Institute, Inc.,
9 Research Committee for HERP Studies; pp. 11-84.
10
11 Iwai, K.; Udagawa, T.; Yamagishi, M.; Yamada, H. (1986) Long-term inhalation studies of diesel exhaust on
12 F344 SPF rats. Incidence of lung cancer and lymphoma. In: Ishinishi, N.; Koizumi, A.; McClellan,
13 R.O.; Stober, W., eds. Carcinogenic and mutagenic effects of diesel engine exhaust: proceedings of the
14 international satellite symposium on lexicological effects of emissions from diesel engines; July; Tsukuba
15 Science City, Japan. Amsterdam, Holland: Elsevier Science Publishers B. V.; pp. 349-360.
16 (Developments in toxicology and environmental science: v. 13).
17
18 Jones, K. H. (1993) Diesel truck emissions, an unrecognized source of PCDD/PCDF exposure in the United
19 States. Risk Anal. 13: 245-252.
20
21 Karagianes, M. T.; Palmer, R. F.; Busch, R. H. (1981) Effects of inhaled diesel emissions and coal dust in rats.
22 Am. Ind. Hyg. Assoc. J. 42: 382-391.
23
24 Kawabata, Y.; Iwai, K.; Udagawa, T.; Tukagoshi, K.; Higuchi, K. (1986) Effects of diesel soot on unscheduled
25 DNA synthesis of tracheal epithelium and lung tumor formation. In: Ishinishi, N.; Koizumi, A.;
26 McClellan, R. O.; Stober, W., eds. Carcinogenic and mutagenic effects of diesel engine exhaust:
27 proceedings of the international satellite symposium on lexicological effects of emissions from diesel
28 engines; July; Tsukuba Science City, Japan. Amsterdam, The Netherlands: Elsevier Science Publishers
29 B. V.; pp. 213-222. (Developments in toxicologial and environmental science: v. 13).
30
31 Kawabata, Y.; Udagawa, T.; Higuchi, K.; Yamada, H.; Iwai, K. (1994) Early one-year exposure to diesel
32 exhaust causes lung cancer in rals. In: Mohr, U.; Dungworth, D. L.; Mauderly, J. L.; Oberdorster, G.,
33 eds. Toxic and carcinogenic effects of solid particles in the respiratory tracl: [proceedings of the 4lh
34 international inhalation symposium]; March 1993; Hannover, Germany. Washington, DC: International
35 Life Sciences Institute Press; pp. 429-431.
36
37 Kotin, P.; Falk, H. L.; Thomas, M. (1955) Aromatic hydrocarbons: III. presence in the paniculate phase of
38 diesel-engine exhausts and the carcinogenicity of exhausl exlracls. AMA Arch. Ind. Health 11: 113-120.
39
40 Krewski, D.; Crump, K. S.; Farmer, J.; Gaylor, D. W.; Howe, R.; Portier, C.; Salsburg, D.; Sielken, R. L.;
41 Van Ryzin, J. (1983) A comparison of statistical methods for low dose extrapolation utilizing
42 time-lo-lumour daia. Fundam. Appl. Toxicol. 3: 140-160.
43
44 Kuschner, M. (1968) The J. Bums Amberson lecture: the causes of lung cancer. Am. Rev. Respir. Dis.
45 98: 573-590.
46
47 Lee, K. P.; Henry, N. W., Ill; Trochimowicz, H. J.; Reinhardl, C. F. (1986) Pulmonary response lo impaired
48 lung clearance in rats following excessive TiO2 dust deposition. Environ. Res. 41: 144-167.
49
50 Lerchen, M. L.; Wiggins, C. L.; Samet, J. M. (1987) Lung cancer and occupation in New Mexico. JNCI
51 J. Natl. Cancer Inst. 79: 639-645.
52
53 Lloyd, J. W. (1971) Long-term mortality study of steel workers: V. respiratory cancer in coke plant workers.
54 J. Occup. Med. 13: 53-68.
December 1994 11-42 DRAFT-DO NOT QUOTE OR CITE
-------
1 Marklund, S.; Andersson, R.; Tysklind, M.; Rappe, C.; Egeback, K.-E.; Bjorkman, E.; Grigoriadis, V. (1990)
2 Emissions of PCDDs and PCDFs in gasoline and diesel fueled cars. Chemosphere 20: 553-561.
3
4 Martin, J. C.; Daniel, H.; Le Bouffant, L. (1977) Short- and long-term experimental study of the toxicity of coal
5 mine dust and some of its constituents. In: Walton, W. H., ed. Inhaled particles IV, pan 1: proceedings
6 of an international symposium; September 1975; Edinburgh, United Kingdom. Oxford, United Kingdom:
7 Pergamon Press; pp. 361-371.
8
9 Mauderly, J. L.; Jones, R. K.; Griffith, W. C.; Henderson, R. F.; McClellan, R. O. (1987) Diesel exhaust is a
10 pulmonary carcinogen in rats exposed chronically by inhalation. Fundam. Appl. Toxicol. 9: 208-221.
11
12 Mauderly, J. L.; Snipes, M. B.; Barr, E. B.; Bechtold, W. E.; Henderson, R. F.; Mitchell, C. E.; Nikula,
13 K. J.; Thomassen, D. G. (1991) Influence of particle-associated organic compounds on carcinogenicity of
14 diesel exhaust. Presented at: eighth Health Effects Institute annual conference; April; Colorado Springs,
15 CO. Cambridge, MA: Health Effects Institute.
16
17 Mauderly, J. L.; Henderson, R. F.; Nikula, K. J.; Snipes, M. B. (1993) Non-cancer effects of chronic inhalation
18 exposure of animals to solid particles. In: Toxic and carcinogenic effects of solid particles in the
19 respiratory tract—proceedings of the 4th international inhalation symposium; March; Hannover,
20 Germany, p. 42.
21
22 McClellan, R. O. (1986) Health effects of diesel exhaust: a case study in risk assessment. Am. Ind. Hyg. Assoc.
23 J. 47: 1-13.
24
25 McClellan, R. 0.; Cuddihy, R. G.; Griffith, W. C.; Mauderly, J. L. (1989) Integrating diverse data sets to
26 assess the risks of airborne pollutants. In: Mohr, U.; Bates, D. V.; Dungworth, D. L.; Lee, P. N.;
27 McClellan, R. 0.; Roe, F. J. C., eds. Assessment of inhalation hazards: integration and extrapolation
28 using diverse data: [papers from the 2nd international inhalation symposium]; 1989; Hannover, FRG.
29 Berlin, Federal Republic of Germany: Springer-Verlag; pp. 3-22.
30
31 Moolgavkar, S. H.; Knudson, A. G., Jr. (1981) Mutation and cancer: a model for human carcinogens. JNCI
32 J. Natl. Cancer Inst. 66: 1037-1052.
33
34 Morrow, P. E. (1988) Possible mechanisms to explain dust overloading of the lungs. Fundam. Appl. Toxicol.
35 10: 369-384.
36
37 Muhle, H.; Bellmann, B.; Creutzenberg, O.; Dasenbrock, C.; Emst, H.; Kilpper, R.; MacKenzie, J. C.;
38 Morrow, P.; Mohr, U.; Takenaka, S.; Mermelstein, R. (1991) Pulmonary response to toner upon
39 chronic inhalation exposure in rats. Fundam. Appl. Toxicol. 17: 280-299.
40
41 Mumford, J. L.; Chapman, R. S.; Harris, D. B. (1989) Indoor air exposure to coal and wood combustion
42 emissions associated with a high lung cancer rate in Xuan Wei, China. Environ. Int. 15: 315-320.
43
44 National Council on Radiation Protection and Measurements. (1980) Influence of dose and its distribution in time
45 on dose-response relationships for low-LET radiations: recommendations of the National Council on
46 Radiation Protection and Measurements. Washington, DC: National Council on Radiation Protection and
47 Measurement; NCRP report no. 64.
48
49 Nesnow, S.; Evans, C.; Stead, A.; Creason, J.; Slaga, T. J.; Triplett, L. L. (1982) Skin carcinogenesis studies
50 of emission extracts. In: Lewtas, J., ed. Toxicological effects of emissions from diesel engines:
51 proceedings of the Environmental Protection Agency diesel emissions symposium; October 1981;
52 Raleigh, NC. New York, NY: Elsevier Biomedical; pp. 295-320. (Developments in toxicology and
53 environmental science: v. 10).
54
December 1994 ^.43 DRAFT-DO NOT QUOTE OR CITE
-------
1 Nikula, K. J.; Snipes, M. B.; Barr, E. B.; Griffith, W. C.; Henderson, R. F.; Mauderly, J. L. (1994) Influence
2 of particle-associated organic compounds on the carcinogenicity of diesel exhaust. In: Mohr, U.;
3 Dungworth, D. L.; Mauderly, J. L.; Oberdorster, G., eds. Toxic and carcinogenic effects of solid
4 particles in the respiratory tract: [proceedings of the 4th international inhalation symposium]; March
5 1993; Hannover, Germany. Washington, DC: International Life Sciences Institute Press; pp. 565-568.
6
7 Oberdorster, G.; Yu, C. P. (1991) The carcinogenic potential of inhaled diesel exhaust: a particle effect?
8 J. Aerosol Sci. 21(suppl. 1): S397-S401.
9
10 Oehme, M.; Larssen, S.; Brevik, E. M. (1991) Emission factors of PCDD and PCDF for road vehicles obtained
11 by tunnel experiment. Chemosphere 23: 1699-1708.
12
13 Pepelko, W. E.; Peirano, W. B. (1983) Health effects of exposure to diesel engine emissions: a summary of
14 animal studies conducted by the U.S. Environmental Protection Agency's Health Effects Research
15 Laboratories at Cincinnati, Ohio. J. Am. Coll. Toxicol. 2: 253-306.
16
17 Piantadosi, S. (1994) Invited commentary: ecologic biases. Am. J. Epidemiol. 139: 761-764.
18
19 Pierson, W. R.; Brachaczek, W. W. (1976) Paniculate matter associated with vehicles on the road. Warrendale,
20 PA: Society of Automotive Engineers; SAE technical paper no. 760039. SAE Trans. 85: 209-227.
21
22 Portier, C.; Hedges, J. C.; Hoel, D. G. (1986) Age-specific models of mortality and tumor onset for historical
23 control animals in the National Toxicology Program's carcinogenicity experiments. Cancer Res.
24 46: 4372-4378.
25
26 Pott, F.; Heinrich, U. (1987) Dieselmotorabgas und Lungenk auf die Gefahrdung des Menschen. Umwelthygiene
27 (Dusseldorf) 19: 130-167.
28
29 Raffle, P. A. B. (1957) The health of the worker. Br. J. Ind. Med. 14: 73-80.
30
31 Rinsky, R. A.; Young, R. J.; Smith, A. B. (1981) Leukemia in benzene workers. Am. J. Ind. Med. 2: 217-245.
32
33 Rushton, L.; Alderson, M. R.; Nagarajah, C. R. (1983) Epidemiological survey of maintenance workers in
34 London Transport Executive bus garages at Chiswick Works. Br. J. Ind. Med. 40: 340-345.
35
36 Saffiotti, U. et al. (1988) Multifactorial hamster respiratory carcinogenesis: oxide, pyrene and
37 N-methyl-N-nitrosourea (MNU). Presented at: Biology, toxicology and carcinogenesis of respiratory
38 epithelium; November; Albuquerque, NM.
39
40 Sagai, M.; Saito, H.; Ichinose, T.; Kodama, M.; Mori, Y. (1993) Biological effects of diesel exhaust particles.
41 I. In vitro production of superoxide and in vivo toxicity in mouse. Free Radical Biol. Med. 14: 37-47.
42
43 Sanders, C. (1989) [Personal communication].
44
45 Schuetzle, D.; Frazier, J. A. (1986) Factors influencing the emission of vapor and paniculate phase components
46 from diesel engines. In: Ishinishi, N.; Koizumi, A.; McClellan, R. O.; Stober, W., eds. Carcinogenic
47 and mutagenic effects of diesel engine exhaust: proceedings of the international satellite symposium on
48 toxicological effects of emissions from diesel engines; July; Tsukuba Science City, Japan. Amsterdam,
49 The Netherlands: Elsevier Science Publishers B. V.; pp. 41-63. (Developments in toxicology and
50 environmental science: v. 13).
51
52 Schum, M.; Yeh, H.-C. (1980) Theoretical evaluation of aerosol deposition in anatomical models of mammalian
53 lung airways. Bull. Math. Biol. 42: 1-15.
54
December 1994 11-44 DRAFT-DO NOT QUOTE OR CITE
-------
1 Smith, L. R. (1989) Characterization of exhaust emissions from trap-equipped light-duty diesels [final report].
2 Sacramento, CA: California Air Resources Board; contract no. A5-159-32.
3
4 Smith, R. A.; Stayner, L. (1990) An exploratory assessment of the risk of lung cancer associated with exposure
5 to diesel exhaust based on a study with rats [final report]. Cincinnati, OH: National Institute for
6 Occupational Safety and Health, Division of Standards Development and Technology Transfer.
7
8 Steenland, N. K.; Silverman, D. T.; Hornung, R. W. (1990) Case-control study of lung cancer and truck driving
9 in the Teamsters Union. Am. J. Public Health 80: 670-674.
10
11 Stenback, F.; Rowland, J.; Sellakumar, A. (1976) Carcinogenicity of pyrene and dusts in the hamster lung
12 (instilled intratracheally with titanium oxide, aluminum oxide, carbon and ferric oxide). Oncology
13 33: 29-34.
14
15 Stober, W. (1986) Experimental induction of tumors in hamsters, mice and rats after long-term inhalation of
16 filtered and unfiltered diesel engine exhaust. In: Ishinishi, N.; Koizumi, A.; McClellan, R. O.;
17 Stober, W., eds. Carcinogenic and mutagenic effects of diesel engine exhaust: proceedings of the
18 international satellite symposium on toxicological effects of emissions from diesel engines; July; Tsukuba
19 Science City, Japan. Amsterdam, The Netherlands: Elsevier Science Publishers B. V.; pp. 421-439.
20 (Developments in toxicology and environmental science: v. 13).
21
22 Strom, K. (1987) Presented at: HEI diesel symposium; February; Seabrooke Island, SC.
23
24 Sun, J. D.; Wolff, R. K.; Kanapilly, G. M.; McClellan, R. O. (1984) Lung retention and metabolic fate of
25 inhaled benzo[a]pyrene associated with diesel exhaust particles. Toxicol. Appl. Pharmacol. 73: 48-59.
26
27 Swenberg, J. A.; Kerns, W. D.; Mitchell, R. I.; Gralla, E. J.; Pavkov, K. L. (1980) Induction of squamous cell
28 carcinomas of the rat nasal cavity by inhalation exposure to formaldehyde vapor. Cancer Res.
29 40: 3398-3401.
30
31 Tong, H. Y.; Karasek, F. W. (1984) Quantitation of polycyclic aromatic hydrocarbons in diesel exhaust
32 particulate matter by high-performance liquid chromatography fractionation and high-resolution gas
33 chromatography. Anal. Chem. 56: 2129-2134.
34
35 Vostal, J. J. (1986) Factors limiting the evidence for chemical carcinogenicity of diesel emissions in long-term
36 inhalation experiments. In: Ishinishi, N.; Koizumi, A.; McClellan, R. O.; Stober, W., eds. Carcinogenic
37 and mutagenic effects of diesel engine exhaust: proceedings of the international satellite symposium on
38 toxicological effects of emissions from diesel engines; July; Tsukuba Science City, Japan. Amsterdam,
39 The Netherlands: Elsevier Science Publishers B. V.; pp. 381-396. (Developments in toxicology and
40 environmental science: v. 13).
41
42 Waller, R. E. (1981) Trends in lung cancer in London in relation to exposure to diesel fumes. Environ. Int.
43 5: 479-483.
44
45 Williams, R. R.; Stegens, N. L.; Goldsmith, J. R. (1977) Associations of cancer site and type with occupation
46 and industry from the Third National Cancer Survey interview. J. Natl. Cancer Inst. 59: 1147-1185.
47
48 Wojciak, J. F. (1988) Theoretical and experimental analyses of aerosol deposition in the lung: implications for
49 human health effects [PhD thesis]. Rochester, NY: University of Rochester.
50
51 Wolff, R. K.; Griffis, L. C.; Hobbs, C. H.; McClellan, R. O. (1982) Deposition and retention of 0.1 urn
52 67Ga2O3 aggregate aerosols in rats following whole body exposures. Fundam. Appl. Toxicol. 2: 195-200.
December 1994 H_45 DRAFT-DO NOT QUOTE OR CITE
-------
1 Wolff, R. K.; Henderson, R. F.; Snipes, M. B.; Sun, J. D.; Bond, J. A.; Mitchell, C. E.; Mauderly, J. L.;
2 McClellan, R. O. (1986) Lung retention of diesel soot and associated organic compounds.
3 In: Ishinishi, N.; Koizumi, A.; McClellan, R.; Stober, W., eds. Carcinogenic and mutagenic effects of
4 diesel engine exhaust: proceedings of the international satellite symposium on lexicological effects of
5 emissions from diesel engines; July; Tsukuba Science City, Japan. Amsterdam, The Netherlands: Elsevier
6 Science Publishers B. V.; pp. 199-211. (Developments on toxicology and environmental science: v. 13).
7
8 Wong, O.; Morgan, R. W.; Kheifets, L.; Larson, S. R.; Whorton, M. D. (1985) Mortality among members of a
9 heavy construction equipment operators union with potential exposure to diesel exhaust emissions.
10 Br. J. Ind. Med. 42: 435-448.
11
12 Woskie, S. R.; Smith, T. J.; Hammond, S. K.; Schenker, M. B.; Garshick, E.; Speizer, F. E. (1988a)
13 Estimation of the diesel exhaust exposures of railroad workers: I. current exposures. Am. J. Ind. Med.
14 13: 381-394.
15
16 Woskie, S. R.; Smith, T. J.; Hammond, S. K.; Schenker, M. B.; Garschick, E.; Speizer, F. E. (1988b)
17 Estimation of the diesel exhaust exposures of railroad workers: II. national and historical exposures. Am.
18 J. Ind. Med. 13: 395-404.
19
20 Yu, C. P.; Morrow, P. E.; Chan, T. L.; Strom, K. A.; Yoon, K. J. (1987) A nonlinear model of alveolar
21 clearance of insoluble particles from the lung. Inhalation Toxicol. (premier issue): 97-107.
22
23 Yu, C. P.; Yoon, K. J. (1988) Determination of lung doses of diesel exhaust perticulates. Presented at:
24 5th annual conference of Health Effects Institute; Colorado Springs, CO.
25
26 Yu, C. P.; Yoon, K. J.; Chen, Y. K. (1991) Retention modeling of diesel exhaust particles in rats and humans.
27 J. Aerosol Med. 4: 79-115.
December 1994 U-46 DRAFT-DO NOT QUOTE OR CITE
-------
i 12. HEALTH RISK CHARACTERIZATION FOR
2 DIESEL ENGINE EMISSIONS
3
4
5 12.1 INTRODUCTION
6 The purpose of a health risk characterization is to communicate to an audience the
7 nature and extent of possible human health hazards associated with toxic agents, the
8 circumstances under which the hazards may be present and the confidence or uncertainty
9 about the hazards. A risk characterization consists of several elements. The first is a
10 determination of how likely humans are to be affected and if so, how serious are the effects.
11 The second element is a dose-response assessment in which the onset and degree of adverse
12 effect is correlated with exposure, and if necessary scaled to a human. The third element is
13 an estimate of human exposure levels. This is often obtained by a combination of both actual
14 measurements and exposure modeling. The fourth element is an integration of the dose-
15 response data with exposure estimates to describe possible impacts on an exposed population.
16 The final element is an analysis of the uncertainties of both the evaluation and the melding of
17 the information into a story that can be understood by the target audience.
18 The characterization of health hazard potential for diesel exhaust is less straight-forward
19 than for a single toxic agent because it is a complex mixture made up of carbon particles
20 with numerous adsorbed compounds and a vapor phase also consisting of a variety of organic
21 as well as inorganic components. The characterization of human exposure to diesel
22 emissions is crude and incomplete because of the difficulty in estimating ambient exposure
23 levels. Because, in most cases, it is impossible to separate diesel particles from others
24 present in ambient air, exposure estimates must usually be determined indirectly.
25 A combining of the health hazard data and the exposure information for purposes of
26 describing the likelihood and possible magnitude of adverse effects includes uncertainties
27 associated with both risk and exposure estimates.
28
29
December 1994 12-1 DRAFT-DO NOT QUOTE OR CITE
-------
1 12.2 ACUTE EXPOSURE HAZARDS
2 12.2.1 Hazard Identification
3 The effects reported following short-term exposures of experimental animals were, in
4 most cases, limited to the lungs. Pulmonary edema may occur during the first days of
5 exposure. Aggregations of particle laden macrophages in the alveolar regions of the lungs,
6 Type II cell proliferation and thickening of alveolar walls were also common features after
7 several days exposure. In some cases, weight losses were reported and minimal changes in
8 lung function were seen.
9 The primary effects noted in occupationally exposed individuals were symptoms of
10 mucous membrane irritation, headache, lightheadedness, and dizziness. Diesel exhaust odor
11 is considered unpleasant enough to induce psychological effects at high concentrations.
12 Except for decreased expiratory flow rates over the course of a workshift in one study, few
13 changes in pulmonary function were noted.
14
15 12.2.2 Dose Response for Acute Toxicity
16 Because pulmonary effects were seen only in experimental animals at concentrations of
17 several milligrams per cubic meter, they are unlikely to be critical endpoints for acutely
18 exposed humans. The probable critical effects in humans (odor, headache, and mucous
19 membrane irritation) were shown to have thresholds for diesel exhaust diluted as much as
20 475-fold with clean air (i.e., about 200 jig/m3 particle concentration). Although a
21 no-observable-adverse-effect (NOAEL) for acute exposure was not formally determined, it
22 appears that concentrations greater than 200 /*g/m3 may be noxious. For sensitive
23 individuals, even lower concentrations may induce adverse effects.
24
25
26 12.3 CHRONIC NONCARCINOGENIC EXPOSURE HAZARDS
27 12.3.1 Hazard Identification
28 The ability of diesel exhaust to induce adverse human health effects other than cancer
29 was assessed by evaluating both subchronic and chronic animal bioassays as well as human
30 epidemiological data. Several epidemiological studies are available to aid in evaluating the
31 effects of chronic exposure on occupationally exposed workers. These epidemiology studies
December 1994 12-2 DRAFT-DO NOT QUOTE OR CITE
-------
1 show an effect on pulmonary function, but are not definitive for diesel exposure, because of
2 possible coexposure to other pollutants.
3 An extensive animal database is available including studies with rats, mice, hamsters,
4 cats, and monkeys. The species with the most extensive database and the one responding at
5 the lowest doses is the rat. The studies are consistent among all species in that the critical
6 target site is the deep lung. In these studies, the common endpoints seen were aggregations
7 of particle laden macrophages in the alveolar regions. This was accompanied by focal
8 thickening of the alveolar walls, replacement of Type I alveolar cells by Type II cells, and
9 fibrosis. These changes were most evident adjacent to the aggregations of macrophages.
10 The severity of these inflammatory responses were directly related to exposure levels.
11
12 12.3.2 Dose Response for Chronic Toxicity
13 12.3.2.1 Selection of Dose-Response Data
14 Two studies with rats were selected for determining safe levels for chronic exposure.
15 In these two studies, in which exposure concentrations ranged from 0.11 to 7.0 mg/m3
16 particulate matter, a large number of endpoints were measured. The chronic animal data
17 were consistent in identifying a threshold region, the concentration range separating the
18 lowest concentrations at which adverse effects are observed in an experimental study, and the
19 highest tested concentrations at which effects were not observed. Because of variations in
20 experimental design and exposure regime, the two studies selected, as well as supporting
21 ones examining chronic diesel emission exposures, are not directly comparable. Despite
22 these differences, the consistency in identifying the threshold region was remarkable, adding
23 to the confidence in the no-observed-adverse-effect level (NOAEL) that was used in
24 derivation of the RfC.
25
26 12.3.2.2 Dose Measure
27 The dose measure is considered to be the carbon particle core. (See the discussion on
28 cancer quantitation [Section 12.4.2.2]).
29
30
December 1994 !2_3 DRAFT-DO NOT QUOTE OR CITE
-------
1 12.3.2.3 Dose Equivalence Across Species
2 The dosimetry model described in the cancer section is used (Section 12.4.2.3).
3 As in the cancer risk assessment, it was assumed that equivalent sensitivity occurs across
4 species when lung dose is expressed as mass of particles per unit surface area. See
5 Chapter 11 for a discussion of this issue.
6
7 12.3.2.4 Inhalation Reference Concentration Derivation
8 The approach used for evaluation of the concentration-response information for diesel
9 emissions was the derivation of an inhalation reference concentration (RfC). The RfC is
10 defined as an estimate (with uncertainty spanning perhaps and order of magnitude) of a
11 continuous inhalation exposure to the human population (including sensitive subgroups) that
12 is likely to be without appreciable risks of deleterious effects during a lifetime. Although
13 data derived from human exposures are preferred for derivation of an RfC, the available
14 human diesel exhaust data are inadequate due to confounding, coexposure to other particles,
15 and inadequate exposure measurement.
16 After determining the no observable adverse effect level (NOAEL) for the critical
17 effect(s), adjusting dose from animals to humans and from experimental exposure durations
18 to continuous exposure, the modified NOAEL is divided by an uncertainty factor. The size
19 of this factor is determined by the exposure duration, quality, and completeness of the
20 database.
21 The highest NOAEL observed in the principal rat studies was 0.46 mg/m3. This
22 concentration was reduced to 0.15 mg/m3 to produce an equivalent dose to the lungs of
23 humans exposed continuously, then divided by an uncertainty factor of 30. The uncertainty
24 factor was applied to account for variations in sensitivity among humans and between humans
25 and rats. After these adjustments the RfC of 5 /zg/m3 was derived.
26
27 12.3.2.5 Reasonableness and Utility of the Inhalation Reference Concentration
28 There is considerable evidence that the RfC is protective against adverse effects in
29 human populations without large preexisting lung burdens. The critical endpoints and
30 threshold levels agreed quite well among studies, despite differences in exposure regimes and
31 experimental designs. Because adverse effects were noted in other organs only at
December 1994 12-4 DRAFT-DO NOT QUOTE OR CITE
-------
1 considerably higher concentrations, it is unlikely that further studies will result in lower
2 effect levels. Furthermore, only minimal pathological effects were reported in monkeys, a
3 species more closely related to humans, exposed to concentrations nearly fourfold greater
4 than the NOAEL in rats.
5 On the other hand, for individuals with preexisting lung burdens, the inference of
6 protection at the RfC concentration may not be correct. The possibility exists that additional
7 particle deposition resulting from concentrations at the RfC may exceed the no-observable-
8 adverse-level. This concern cannot be further defined at present. There is little data
9 concerning existing particle lung burdens in the general population or the degree of adaptivity
10 of diesel exhaust particles with other ambient paniculate matter.
11
12
13 12.4 CARCINOGENIC EXPOSURE HAZARDS
14 12.4.1 Hazard Identification
15 As previously mentioned diesel exhaust contains a number of components that have
16 been shown to be carcinogenic in experimental animals. Extensive studies with several
17 strains of Salmonella typhimurium have unequivocally demonstrated mutagenic activity of
18 diesel exhaust particle extracts, both with and without rat liver S9 activation. Positive results
19 also have been reported in a variety of mammalian cell tests. Sister chromatic exchanges are
20 increased in a variety of studies with diesel particles and diesel particle extracts. Finally,
21 DNA abducts were increased in lung epithelial cells of rats exposed to whole diesel exhaust.
22 Overall, the evidence for lung cell DNA damage (e.g., genotoxicity) by diesel exhaust is
23 quite strong.
24 Evaluation of the mutagenic activity of diesel exhaust particles stripped of organics has
25 not been carried out. Presently available mutagenicity assay methods may not be appropriate
26 because particle effects are thought to be mediated indirectly through release of various
27 factors by particle laden macrophages and other phagocytic cells. An increase in DNA
28 abducts, however, was seen following exposure to carbon black, which is similar to the
29 diesel particle stripped of organics.
30 Chronic diesel exhaust inhalation exposure has resulted in the induction of lung cancer
31 in several long-term studies in rats and mice. Respiratory tract tumors also were induced
December 1994 12-5 DRAFT-DO NOT QUOTE OR CITE
-------
1 following instillation of either organic extracts of diesel exhaust particles, or particles
2 "stripped" of organics, into the lungs of rats. Dermal tumors were induced by skin painting
3 with diesel particle extracts. Overall, these studies provided a broad spectrum of positive
4 evidence for the carcinogenicity of diesel exhaust in experimental animals. The weight-of-
5 evidence for animal data according to EPA's 1986 Carcinogenicity Risk Assessment
6 Guidelines is "sufficient" (Federal Register, 1986).
7 Increased lung cancer mortality was reported in a number of epidemiology studies. The
8 usefulness of the epidemiology data in assessing cancer risk is reduced because of
9 methodological limitations such as small sample size, short follow up, or lack of adequate
10 adjustment for confounding factors. Nevertheless, some of the more recent studies,
11 especially those of Garshick et al. (1987, 1988) were able to overcome most of these
12 deficiencies. Collectively, the epidemiology studies show evidence of an association between
13 inhalation of diesel exhaust and lung cancer in humans. Although the evidence for
14 carcinogenicity in humans was in most cases positive, it is judged to be "limited" according
15 to EPA's weight-evidence-guidelines, because the observed increases in risk were quite low
16 and the influence of confounding factors could not be completely accounted for.
17 Considering both the animal and human data together, diesel exhaust has been classified
18 as a probable human carcinogen and placed in EPA weight-of-evidence-category Bl. This
19 classification is strongly supported by positive data from short-term tests and from data
20 regarding chemical and physical composition.
21
22 12.4.2 Methods for Determining Dose Response
23 In order to develop a quantitative risk estimate using animal bioassays, the following
24 information is required: (1) response data from chronic bioassays of acceptable quality,
25 (2) a suitable measure of dose, (3) a deposition and retention model for estimating dose
26 equivalence between animals and humans, and (4) availability of a suitable high to low dose
27 extrapolation model.
28
29 12.4.2.1 Selection of Dose Response Data
30 Human data are preferable for developing risk estimates. However, for reasons
31 described in Chapter 11, use of the human data are, in this case considered to be inadequate
December 1994 12-6 DRAFT-DO NOT QUOTE OR CITE
-------
1 for this purpose. First of all, the relative risk ratios for the human epidemiology studies are
2 generally only slightly greater than one. Small errors in the adjustment for possible
3 confounding factors, especially smoking, could result in a large percentage change in the
4 relative risk. Secondly, most exposure estimates are indirect because it is virtually
5 impossible to separate diesel paniculate matter from other particles present in the ambient
6 environment or the workplace. Finally, although considerable efforts have been made to
7 estimate diesel exhaust exposure for occupationally exposed groups, most of these
8 measurements are quite recent. Because of long latency periods for tumor development,
9 exposure levels 20 to 30 years previously are necessary for accurate dose-response estimates.
10 Because of improvements in engine efficiency and workplace ventilation, past exposure levels
11 can, in most cases, only be guessed at. Finally, an attempt was made to use the Garshick et
12 al. (1987) railroad worker study to develop a unit risk estimate. However, attempts to relate
13 increasing duration or intensity of exposure to increasing response rates were unsuccessful.
14 On the other hand, several well-controlled and designed chronic exposure studies using
15 rats have been carried out. Rats were the only animal species producing definitive
16 carcinogenic effects. Despite some uncertainties in extrapolating results of animal studies to
17 humans, the rat data was considered to be a preferable surrogate. The rats were generally
18 exposed at higher levels than even occupationally exposed humans, although there was some
19 overlap. Particle concentrations in the animal studies ranged from 0.1 to as great as
20 12 mg/m3, whereas present exposure levels for railroad workers only range up to about
21 0.2 to 0.3 mg/m3. Earlier exposures, however, probably exceeded 1 mg/m3 for certain
22 groups such as workers in railroad engine repair facilities.
23 Chronic bioassays with several species have been reported. Three studies, Brightwell
24 et al. (1986), Ishinishi et al. (1986), and Mauderly et al. (1987), were considered to be well
25 suited for risk estimation analysis. These three studies were all of 2 years duration or
26 longer, all used Fischer 344 rats, they collectively included about a 50-fold range of
27 exposures and all showed lung tumor induction.
28
29 12.4.2.2 Dose Measure
30 Three measures of dose were considered: (1) the vapor phase, (2) the particle adsorbed
31 compounds (i.e., extractable organics), and (3) the carbon core of the particle. The vapor
December 1994 12-7 DRAFT-DO NOT QUOTE OR CITE
-------
1 phase was not selected for two reasons. First of all, increases in lung cancer incidence in
2 animals were not reported following exposure to the vapor phase alone. Although the
3 possibility of positive responses at higher doses cannot be ruled out, the lack of positive data
4 precludes development of a unit risk estimate based upon this component. Secondly,
5 although the vapor phase contains some potentially carcinogenic agents, their site of action
6 has been other than the deep lung, the primary target for diesel effects.
7 The particle adsorbed organic fraction does contain compounds known to induce lung
8 cancer in experimental animals. This fraction was considered, but not ultimately selected,
9 because the concentration of these compounds were very low and unlikely to induce the
10 degree of carcinogenic response seen. For example, in one study (Tong and Karasek, 1984),
11 the total concentration of 43 PAHs measured was only about 5 /xg/mg of paniculate matter.
12 Furthermore only a few of these are known to be carcinogenic. Among the carcinogenic
13 ones, the concentration of B[a]P was less than 0.1 /ig/mg of paniculate matter.
14 Nevertheless, possible effects of the adsorbed organics cannot be totally discounted and they
15 are considered likely to contribute to the tumorigenic responses.
16 The carbon core of the particle was selected as the preferred dose measure because it
17 appears capable of not only inducing carcinogenic effects but accounting for the total animal
18 response seen. Diesel particles stripped of adorbed organics induced lung cancer following
19 intratracheal instillation. Inhalation of carbon black, which is similar to the carbon core of
20 the diesel particle, resulted in about the same degree of carcinogenic response as whole diesel
21 exhaust.
22
23 12.4.2.3 Dose Equivalence Across Species
24 A dosimetry model was used to account for species differences in respiration rates;
25 particle deposition efficiency; particle clearance rates, both between species and from high to
26 low dose; transport of particles to lung associated lymph nodes; and lung surface area.
27 Owing to the availability of much research on dosimetry in the rat and human lung, the use
28 of such a model was preferable to using other types of assumptions to accomplish the same
29 purpose. The dosimetry was based upon the assumption that an equivalent particle
30 concentration per unit of lung surface area will result in an equivalent tumorigenic response
December 1994 12-8 DRAFT-DO NOT QUOTE OR CITE
-------
1 in rats and humans. The model's parameters were largely based upon experimentally derived
2 values rather than assumptions. Details of the model are shown in Appendix D.
3 Particle clearance is a particularly important issue as it influences the length of time the
4 diesel exhaust residue resides at the target tissue site. Rates of particle clearance from the
5 deep lung differ among species, with rats having normal clearance half-times of about 2 mo
6 versus 1 year in humans. This will result in a larger buildup of particles in human than rat
7 lungs. Offsetting this difference to some extent is the finding that clearance is slowed .at
8 some of the higher concentrations used in the rat bioassays.
9 The particle associated organics, on the other hand, are eluted sufficiently rapidly that
10 lung burdens of this component are only marginally influenced by particle clearance rates.
11 Vapor phase components are also relatively unaffected by lung burden. Thus, depending
12 upon which components of diesel exhaust are considered primarily responsible for lung
13 cancer induction, differing estimates of human target tissue dose, and, thereby, estimates of
14 risk will be obtained.
15
16 12.4.2.4 High-to-Low-Dose Risk Extrapolation
17 The selection of a model for low-dose extrapolation depends upon the availability of
18 information concerning mechanisms of toxic action. The EPA's 1986 carcinogenicity
19 guidelines (Federal Register, 1986) provide the flexibility in model selection, if available
20 information about mechanisms is sufficient to guide model selection. In the absence of such
21 information, the guidelines recommend a default nonthreshold curve fitting model in the
22 observed data range that has a linear component in the low-extrapolated-dose range.
23 It is possible that lung cancer is primarily the consequence of cytotoxity and subsequent
24 cell proliferation induced by secretions from particle laden macrophages. The likelihood
25 exists, however, that particle adsorbed organics or even vapor phase compounds will induce
26 genetic alteration as well as interact in the proliferative phase. Because of the complexities
27 of the process, uncertainties concerning the actual mechanisms and lack of quantitative data
28 concerning interrelationships of the components, a departure from a linearized model would
29 be arbitrary. Unit risks were therefore calculated based on two different models, both of
30 which are linearized in the low-dose range.
31
December 1994 12-9 DRAFT-DO NOT QUOTE OR CITE
-------
1 Linearized Multistage Model
2 This model is a default choice that is used if another model has not been shown to be
3 more appropriate. The LMS cannot test hypotheses about possible mechanisms of action, it
4 is a curve-fitting model, and its main purpose is to provide upper bound low-dose risk
5 estimates when it can be plausibly assumed that the dose-response is operating in a linear
6 manner below the observable response range. Both malignant as well as combined malignant
7 plus benign tumor data were modeled, as would normally be the case.
8
9 Alternative Biologically Based Model
10 This model was developed to account for the initiating and proliferative effects of both
11 the particle adsorbed organic matter as well as the inorganic particle core. The model has
12 the following properties.
13 1. Accounts for possible effects of both the carbon particle and its associated organics.
14 2. Allows evaluation of the contribution of both the carbon particles and organics to
15 tumor induction.
16 3. Allows for changing of parameters with increasing lung burdens.
17 4. Assumes that cell proliferation and tumor induction are stochastic. For instance, it
18 is not appropriate to assume that all cells divide at the same rate.
19
20 This model, which is illustrated in detail in Appendix C, has the flexibility to consider
21 the role of the initiating properties of both the paniculate matter and the organic fraction as
22 well as possible proliferative effects of the carbon fraction. The possible proliferative effects
23 of organics is not covered in this model because the inert particles alone seem to be able to
24 produce a response similar to whole diesel exhaust, whereas the low concentration of toxic
25 organics is unlikely to significantly increase cell death and proliferation.
26 Most of the data needed to fit the model parameters are lacking, though some can be
27 approximated by curve fitting the model to the observed tumor incidence and mortality data
28 in the experimental exposure range for the rat. This results in an increased uncertainty of
29 risk estimates because of the lack of data at low doses. The model, nevertheless, provides an
30 opportunity to test hypotheses about the relative role of initiation and proliferation and
December 1994 12-10 DRAFT-DO NOT QUOTE OR CITE
-------
1 organics versus particles in risk estimation. The model in its present form also provides an
2 upper bound estimate of risk.
3
4 12.4.3 Results of Dose-Response Calculations
5 12.4.3.1 Results Using the Linearized Multistage Model
6 The results are summarized in Table 12-1. The LMS model was applied to three rat
7 bioassay data sets (Brightwell et al., 1986; Ishinishi et al., Mauderly et al., 1987). The unit
8 risk estimates calculated from these three studies varied from 1.6 to 7.1 x lO'^g/m3)'1
9 with a geometric mean for the three studies of 3.4 x 10"5(/ig/m3)"'. The unit risks are the
10 estimated 95% upper confidence limits of the risk from continuous lifetime exposure to 1
11 ^g/m3 of diesel exhaust paniculate matter. The upper 95 % confidence limit means that the
12 true risk, which cannot be defined, may be less but is unlikely to be more than the calculated
13 value.
14
15
TABLE 12-1. UNIT RISK ESTIMATES PER MICROGRAMS PER CUBIC METER
OF DIESEL EXHAUST
Reference
Mauderly et al. (1987)
Ishinishi et al. (1986)
Brightwell et al. (1986)
Geometric mean of above
Mauderly et al. (1987)
Mauderly et al. (1987)
Mauderly et al. (1987)
Model Used
LMS
LMS
LMS
LMS
LMSa
AMa
AMa
95% Upper Bound of Risk
3.4 X 10-5
1.6 x 10'5
7.1 x 10'5
3.4 x 10'5
1.7 x ID'5
1.7 x 10'5
8.2 X l(T6b
"Using malignant tumors only.
bMaximum likelihood estimate of risk.
LMS = Linearized multistage.
AM = Alternative model.
MLE = Maximum likelihood estimate.
December 1994 12-11 DRAFT-DO NOT QUOTE OR CITE
-------
1 12.4.3.2 Results Using the Alternative Model
2 The data from the Mauderly et al. (1987) study were applied to the alternative model.
3 These data are useful because they contain information on natural mortality and serial
4 sacrifice of animals, valuable for estimating tumor latency. The model parameters were
5 based upon malignant tumors, rather than all tumors as was done in the LMS method. This
6 was necessary in order to utilize mortality data. In order to compare the results with the
7 LMS model, a unit risk was calculated using the LMS model applied to only the malignant
8 tumor data from the Mauderly et al. (1987) study. The results were identical,
9 1.7 X 10~5 (jug/m3)"1, when rounded. This was not unexpected, however, because curve
10 fitting and linearization may be controlling results more than the model form. The maximum
11 likelihood estimate (MLE) was 8.2 x 10'6 (/xg/m3)'1.
12 The effects of varying model coefficients are shown in Appendix C. The model shows
13 that tumor response is very sensitive to an increase in proliferation rate of the initiated cell
14 population and that a small change in proliferation rate will likely cause a disproportionately
15 large change in tumorigenicity and thus risk. Results from the model also suggest that if
16 neither the organics nor particles have an initiating effect, then the resulting risk is
17 significantly less compared to a case where both initiation and proliferation are actively
18 induced by the organic and particle fractions.
19
20 12.4.4 Discussion of Confidence in the Upper Bound Risk Estimates
21 The risk estimates derived are based upon well-designed, well-conducted, and
22 adequately reported rat studies, covering a wide range of doses. The use of a species-
23 specific dosimetry model allows a more accurate estimation of equivalent lung burdens.
24 Finally, epidemiology studies indicate that humans are susceptible to diesel exhaust-induced
25 cancer and that the primary target organ is the same in both rats and humans. Similar target
26 sites increase the chance that cancer potency will not differ greatly between species.
27 Nevertheless, a variety of uncertainties are present, in the estimation of risk from
28 animal data, even at doses in the observable range. Several of these involve extrapolation
29 across species and are common to most risk assessments involving use of animal data. Even
30 though the dosimetry model can provide a plausible, if not reasonable dose extrapolation
31 across species, the possibility of species differences in target site sensitivity exists. In some
December 1994 12-12 DRAFT-DO NOT QUOTE OR CITE
-------
1 cases, this can be reduced by the availability of information regarding metabolic pathways,
2 activation rates, etc. Such data are quite limited for diesel exhaust because of the complexity
3 of the mixture. Human epidemiology data, nevertheless, suggest, at least qualitatively, that
4 both the rat and human are more comparable in sensitivity than grossly different to diesel
5 exhaust-induced tumors.
6 The unit risk estimates for both the LMS and alternative model were derived using
7 concentration of diesel paniculate matter per unit lung surface area as the dosimeter.
8 Although this means of estimating dose is not commonly used in EPA risk assessments, it
9 was considered to be the most plausible assumption because the diesel exhaust-induced lung
10 tumors have been reported to arise from epithelial tissue lining the lung. Although such a
11 dosimetric adjustment was considered to be the most reasonable, because of species
12 differences in total numbers of lung cells, along with the possibility that some tumors may
13 arise from other cells, a degree of uncertainty still remains regarding adoption of this
14 parameter.
15 Unlike most previous assessments, an adjustment for species differences in metabolic
16 rate was not made. Such an adjustment is based on the assumption that because of slower
17 rates of metabolism with accompanying slower rates of detoxification, and/or repair
18 mechanisms the effective dose in humans is less, per unit body weight, than in the smaller
19 mammal. In the case of particles, however, it is believed that the primary cause of cancer
20 induction is not a direct effect of xenobiotics upon lung epithelial cells. Rather, it is the
21 result of various mediators secreted by macrophages, which diffuse to the epithelial cells,
22 following ingestion of particles. The greater predicted sensitivity of human lung epithelial
23 cells, due to lower rates of detoxification, is likely to be offset by slower production of these
24 transforming agents. Such an assumption, although reasonable, is yet unproven.
25 Possible errors in selecting the fraction of exhaust used as the dosimeter also adds
26 some uncertainty to the unit risk estimates. The organic fraction cannot be totally dismissed
27 from playing a role since it does include a variety of carcinogenic compounds. Substituting
28 particle adsorbed organics into the dosimetry model, however, would result in a less than
29 twofold change in the unit risk estimates. Moreover, in an earlier EPA-sponsored effort, a
30 unit risk estimate using in vitro and mouse skin tumor data from particle extracts and based
31 upon the comparative potency method was nearly identical to the present one
December 1994 12-13 DRAFT-DO NOT QUOTE OR CITE
-------
1 (see Table 11-1). Although the agreement may be fortuitous, nevertheless, it indicates that
2 selection of particles as the dosimeter is not likely to be a large source of error.
3 A prime area of uncertainty is the shape of the dose-response curve. For many organic
4 agents, it is sometimes reasonable to assume a straight line extrapolation to very low doses.
5 In the case of particles, cell proliferation may be related to particle overload and the
6 consequent secretion of cancer-inducing mediators by macrophages. This raises the
7 possibility of a change of shape in the dose-response curve at low doses. The relationship
8 between macrophage particle burden and activation, however, is quite complex. The
9 secretion of various mediators by the individual macrophages may be (1) proportional to
10 particle load; (2) disproportional, but with no threshold; or (3) without a threshold. The
11 particles, moreover are unevenly distributed among macrophages; thus, some macrophages
12 may be overloaded even at low doses. Finally, it is not known if the target organ response
13 to the mediators is linear or nonlinear, until additional data show otherwise (Section 12.6.2).
14 Limited positive cancer data from epidemiology studies do provide some confidence for
15 extrapolation to concentrations at least as low as those that can occur during occupational
16 exposure. In one such study, diesel particle concentrations were about an order of magnitude
17 lower than those resulting in detectable carcinogenicity in the animal studies. However,
18 these concentrations are still about an order of magnitude or more greater than found in
19 heavily travelled urban corridors.
20 Both the LMS and the alternative model may misrepresent risk in individuals with
21 preexisting lung burdens of particles. Assuming that the primary cause of lung cancer is
22 related to lung particle burdens, then the presence of all paniculate matter in the lungs would
23 have a cumulative effect upon clearance, although not all particles would be expected to be
24 equivalent in effects upon lung clearance or toxicity. At least some particles are present in
25 the lungs of all individuals. Certain subgroups (e.g., smokers and individuals occupationally
26 exposed to dusty environments) may have existing lung burdens that approach those in the
27 experimental animal studies. If lung burdens are sufficient to inhibit particle clearance, the
28 dose-response curve for diesel carcinogenicity in such individuals may show upward
29 curvelinearity and the risk estimation discussed in this report may even be an underestimate
30 of risk for such individuals. The resolution of such issues is a research need.
December 1994 12-14 DRAFT-DO NOT QUOTE OR CITE
-------
1 In summary, upper bound unit risk estimates were derived using chronic bioassay data
2 from Fischer 344 rats. Equivalent rat/human lung burdens were estimated using a dosimetry
3 model accounting for respiration rates, deposition efficiency, particle clearance rates, and
4 transport to lymph nodes. Dose was based upon concentration of diesel particles per unit of
5 lung surface area. Low-dose extrapolation was carried out using either a LMS model or an
6 alternative low-dose extrapolation model that describes the possible contribution of both
7 initiation and cell proliferation in the cancer risk estimation. Confirming information about
8 mechanism of action at low doses is not available, and, thus, neither modeling approach has
9 any overriding merit in terms of providing a better estimate of risk.
10 The unit risk estimates from the two extrapolation models are upper bound values, each
11 having large uncertainties as to their proximity to the true risk. Although the values are
12 nearly identical for one rat study, this only means that within the realm of upper bound
13 estimating there is relative agreement. Confidence in the upper bound estimates is
14 reasonably good in the observable range, but is much less in the extrapolated range because
15 of the complex biological issues involved. Based upon animal data, it appears that risk may
16 decrease more rapidly than dose at low concentrations. On the other hand, lung cancer is
17 apparently induced in humans by exposure to diesel exhaust at concentrations several-fold
18 less (but still greater than ambient levels) than effective doses in the animal experiments.
19 If true this may reflect greater sensitivity due to existing lung burdens of other paniculate
20 matter, confounding factors, or other factors including greater sensitivity to diesel exhaust.
21
22
23 12.5 EXPOSURE ESTIMATES
24 12.5.1 Methodology
25 The most recent estimates of annual average concentrations to diesel exhaust paniculate
26 matter (DPM) were published in Chapter 9 of EPA's Motor Vehicle-Related Air Toxics
27 Study (U.S. Environmental Protection Agency, 1993). It is necessary to appreciate that
28 diesel emissions per se are not easily distinguishable from other suspended paniculate matter
29 present in ambient air. Thus, the focus of estimating or measuring diesel exposure is directly
30 relatable to the measurement of airborne paniculate matter. The U.S. Environmental
31 Protection Agency (1993) study used two approaches to generate exposure estimates. In the
December 1994 12-i5 DRAFT-DO NOT QUOTE OR CITE
-------
1 first one (DPM), national fleet average emission factors for 1988 were multiplied by the
2 urban and rural grams-per-mile and micrograms-per-cubic-meter conversion factors obtained
3 from EPA's hazardous air pollution exposure model (HAP-EM:1988). Based on this
4 approach, urban annual average concentrations were estimated to be 2 and 1.2 ng/m3 during
5 1990 and 1995. Rural exposure estimates for these two dates were 1.1 and 0.6 ^g/m3,
6 whereas nationwide estimates for these two dates were 1.8 and 1.1 jig/m3, respectively. The
7 nationwide estimate is a population-weighted average between rural and urban levels.
8 In the second method, using ambient monitoring data, total suspended paniculate
9 matter (TSP) for 1990 was determined to equal 48 pig/m3. This can be multiplied by the
10 percent contribution of diesel paniculate matter to TSP that is estimated to be 5.12% as
11 derived from the ratio of 3.5 x 105 metric tons/year diesel emissions to TSP of
12 7.5 X 106 metric tons/year. Five percent of 48 /tg/m3 is equal to 2.2 /itg/m3. Adjustments
13 were then made for times spent in various microenvironments including indoors. This
14 resulted in an integrated exposure estimate of 1.5 ng/m3. The two methods therefore
15 produce approximately comparable results.
16
17 12.5.2 Confidence in Exposure Estimates
18 The average DPM concentrations are estimated using models and require many
19 assumptions to evaluate uncertain input variables. Calculating an upper confidence limit on
20 the average concentration is therefore difficult. The U.S. Environmental Protection Agency
21 (1993) study estimates have no companion numerical estimate of uncertainties. Qualitatively,
22 one can identify some of the factors that contribute to the uncertainty of the estimates. The
23 model used for estimating annual average exposure is based on carbon monoxide (CO) as a
24 surrogate for DPM and motor vehicle emissions in general. In fact, almost all estimates of
25 ambient DPM are indirect due to difficulties in separating DPM from TSP. Another
26 limitation is that the fixed site CO monitoring data used in the model were not adjusted to
27 account for non-motor vehicle sources of CO, because motor vehicles are thought to be the
28 predominant source of CO in urban areas. This assumption may lead to overestimates of
29 exposure to diesel exhaust paniculate matter. On the other hand, the population
30 classification scheme in the model was not intended to account for groups of people who are
December 1994 12-16 DRAFT-DO NOT QUOTE OR CITE
-------
1 both highly exposed and few in number (e.g., toll booth attendants). This may result in an
2 underestimate of exposure for some of the highest exposure groups.
3 Exposure estimates based upon the fraction of diesel paniculate matter in TSP have
4 uncertainties relating output of both diesel exhaust as well as TSP. The diesel-related
5 uncertainties relate to the total number of vehicle miles travelled (VMT), the relative number
6 of VMT in each category of vehicles (automobiles, light duty trucks, etc.), and estimates of
7 particle emission rates. For the years 1988 through 1991, emission rates were based upon
8 adherence to emission standards. Out-of tune or faulty engines may produce considerably
9 more paniculate matter than predicted by test results or standards. Also, certification test
10 results are not quantitatively extrapolable to the variety of vehicular driving and fuel
11 conditions.
12 There is limited information available regarding highly exposed populations. It is
13 known that they exist, but data specific to their exposure is quite limited. The Motor
14 Vehicles Manufacturing Association carried out a detailed analysis of air pollution from
15 diesel engine emissions in the City of Los Angeles (Sienicki and Mago, 1992). They derived
16 a mean estimated concentration of DPM for 1995 of 2.7 /ixg/m3. This is somewhat less than
17 an earlier estimate of 4.4 /^g/m3 (U.S. Environmental Protection Agency, 1983). Differences
18 were primarily due to lower estimates of particle emission rates by Sienicki and Mago.
19 McClellan (1986) estimated that workers on urban freeways and individuals in urban
20 street canyons may be exposed to 15 /xg/m3 of DPM. The accuracy of this estimate,
21 however, is uncertain because it was based upon data from the early 1980s when the mix of
22 vehicles, vehicle miles travelled, and emission standards were different from the present.
23 There was also no estimate of the number of individuals exposed at these higher
24 concentrations.
25 For convenience, Table 12-2 lists the previously discussed estimates. The estimates
26 vary depending upon what assumptions were made, and it is not clear which are better and
27 for what reason.
28
December 1994 12-i7 DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 12-2. ESTIMATED ANNUAL AMBIENT CONCENTRATIONS OF DIESEL
EXHAUST PARTICULATE MATTER3
Year
1986
1990
1995
U.S.
Environmental
Protection
Agency (1993)
Method 1
Rural
—
1.1
0.6
U.S.
Environmental
Protection
Agency (1993)
Method 1
Urban
—
2.0
1.2
U.S.
Environmental
Protection
Agency (1993)
Method 1
Nationwide
—
1.8
1.1
U.S.
Environmental
Protection
Agency (1993)
Method 2
—
1.5
—
MVMA
(1992)
Los Angeles
—
—
2.7
McClellan
(1986)
Highly Exposed
15.0
—
—
"Expressed in micrograms per cubic meter.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
12.6 POPULATION RISKS AND UNCERTAINTIES
With a clear understanding of hazard identification, dose-response evaluation, exposure,
and the attendant uncertainties, one may choose to demonstrate the magnitude of possible
health impacts upon a population by developing estimates of possible disease occurrence.
For carcinogenesis this traditionally is done by combining the cancer unit risk values with the
population exposure. An upper bound estimate of individual risk is thus obtained, and if
population figures are available, the risk can be converted to a hypothetical number of cancer
cases. In the case of noncancer toxicity, estimates of safe concentrations for either acute or
chronic exposure are compared with the range of estimated exposure levels to determine the
degree of exposure above safe levels, the duration of exposure above safe levels, or both.
12.6.1 Population Risks for the Induction of Noncancer Toxicity
12.6.1.1 Population Risks for Acute Exposure
Population risks for acute exposure have not been estimated.
12.6.1.2 Population Risks for Chronic Exposure
An RfC of 5 /ig/m3 is recommended based upon a NOAEL of 0.46 mg/m3 in rats
(adjusted to 0.15 mg/m3 equivalent concentration in humans), combined with an uncertainty
factor of 30. The RfC is greater than concentrations estimated to occur under most ambient
conditions. In certain locations, however, such as tunnels, crowded freeways, city street
canyons, etc., the concentrations of diesel exhaust paniculate matter may exceed the RfC
December 1994
12-18
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
during at least some portions of the day. However, the RfC is based on continuous lifetime
exposure, and therefore high "parttime" exposure corridors need to be reevaluated in terms
of long-term average exposures. See Figure 12-1 relating risk to exposure for an overall
perspective.
10*,
to
tr
I
&
CL
x>
">
TO
10-1
I RfC
f
V
1
Yfc
uionai Mmoieni Estimates
Highly Exposed Ambient Estimates
1 2 3 4 5 6 7 8910 15 20 30 40
Diesel Paniculate Matter Concentration (ng/m3)
Figure 12-1. Relationship of exposure estimates and risk-specific doses.
December 1994
12-19 DRAFT-DO NOT QUOTE OR CITE
-------
l 12.6.2 Population Risks for Induction of Cancer
2 With a clear understanding of hazard identification, dose-response evaluation, exposure,
3 and the attendant uncertainties, one may choose to demonstrate the magnitude of possible
4 health impacts upon a population by developing estimates of disease occurrence. This
5 traditionally is done by combining the cancer risk values with the population exposure to
6 obtain an estimate of individual risk, or if population figures are available, converting the
7 risk number to an upper bound estimate of cancer cases.
8 The usefulness of the cancer unit risk estimates in characterizing population risks must
9 be considered in light of some specific information that is unique to the diesel issue.
10 Confidence is lacking that these risk estimates are protective for some subgroups. Yet for
11 other subgroups risk may be overestimated. Fortunately, information about this is more
12 obvious for diesel exhaust than for other agents. The true risk at low doses may be less than
13 predicted by a linearized low-dose extrapolation model because the proliferative effects of
14 particles is no longer apparent at some point in the low-dose range. A threshold for
15 tumorigenic responses is unlikely, however, because of the initiating properties of chemicals
16 present in the organic fraction of diesel exhaust.
17 Although, for any population, the true risk could be lower for reasons stated above,
18 some circumstances may result in higher risks for certain subgroups. There is concern for
19 individuals with existing lung particle burdens (i.e., smokers and those employed in a
20 number of occupations). Because such individuals would be expected to have a slowed
21 particle clearance rate if lung burdens exceed a threshold, this would infer a steepening of
22 the dose-response curve and a higher risk for diesel exhaust exposure. Greater exposure will
23 occur in certain other subgroups such as laborers or endurance athletes due to increased
24 respiration rates. A variety of metabolic factors (i.e., impaired ability to repair DNA) may
25 exist in some individuals. For comparative purposes these risk estimates share some
26 consistency of approach with many other chemicals. An improved dose-response-based risk
27 characterization, however, cannot be pursued until a better understanding of mechanisms is
28 achieved and data gaps are addressed.
29 Estimates of population exposures are available but their accuracy in representing true
30 exposure likewise cannot be described at this time. Because of uncertainties regarding both
31 ambient exposure levels, as well as those relating to low-dose extrapolation, outright
December 1994 12-20 DRAFT-DO NOT QUOTE OR CITE
-------
1 calculations of possible cancer cases or risk levels may be misleading unless one is aware of
2 these uncertainties. Given these cautions, upper bound estimates of possible cancer mortality
3 could be derived that vary from about 2 X 10~5 for rural residents during 1995 to 7 x 10~5
4 for residents of urban areas during 1990. These values are based upon the U.S.
5 Environmental Protection Agency (1993) study average exposure estimates and the geometric
6 mean of unit risk estimates obtained by applying the LMS model to three animal studies (see
7 Table 12.1). Using the McClellan (1986) exposure estimate for urban street canyons of
8 15 /ig/m3 as a highly exposed population, an individual's risk may be as great as 5 X 10"4.
9 It should be emphasized that these numbers, when viewed together, provide some plausible
10 boundaries for cancer or risk. The uncertainties of these risk estimates are qualitatively
11 identifiable because they include uncertainties related to the unit risk estimate as well as
12 exposure levels. For reasons discussed, risk may be overestimated under most ambient
13 conditions for an average person. On the other hand, the actual risk may be greater for
14 individuals performing heavy labor, exercising in polluted areas, occupationally exposed, or
15 with lungs already burdened by large numbers of particles (see Figure 12-1).
16
17 12.6.3 Comparison of Cancer and Noncancer Risk Estimates
18 The relationship between an RfC and a unit risk is not straightfoward. Although the
19 unit risk estimates a finite cancer rate at any exposure concentration, the RfC is a
20 concentration assumed to be without adverse effects even among sensitive segments of the
21 population. At the RfC concentration, an estimated cancer risk of 1.7/10,000 still exists if
22 the assumption is made that all particle and organic fraction mechanisms are operating. Thus
23 concentrations lower than the RfC are necessary to reduce the estimated cancer risk to less
24 than 10'4. However, because no adverse effects, including cell proliferation, are assumed to
25 occur at this concentration, the cancer rate may be less than 10"4 in a population without
26 preexisting lung burdens. On the other hand, for individuals with preexisting lung particle
27 burdens near the threshold for inhibition of clearance, the additional cancer risk may be
28 greater than predicted.
29
30
31
December 1994 12_2i DRAFT-DO NOT QUOTE OR CITE
-------
1 12.7 SUMMARY
2 Diesel exhaust consists of a vapor phase and participate matter component made up of
3 an insoluble carbon core with sulfates and a large variety of organic compounds adsorbed to
4 the surface. Although a number of the particle-associated and vapor phase compounds are
5 known toxins and are likely to contribute to the disease process, the particle fraction is
6 hypothesized to be the most influential in disease causation. The role of particle extractable
7 organics, however, is not ruled out and probably does have some influence on the total
8 response.
9 Animal studies indicate that chronic exposure to diesel exhaust can result in lung
10 cancer, pulmonary pathology, and possibly neurotoxic effects. The primary effects noted
11 acutely are lung damage at high-exposure concentrations.
12 Although human epidemiology studies indicate that diesel exhaust is likely to be
13 carcinogenic, confounding factors cannot be completely ruled out. Diesel exhaust has been
14 classified into cancer weight-of-evidence category Bl and, thus, is considered to be a
15 probable human carcinogen. Noncancer effects were generally limited to minor changes in
16 pulmonary function. Acute exposure resulted in eye irritation and headaches.
17 An upper bound cancer unit risk estimate of 3.4 x lO'^jUg/m3)"1 is recommended.
18 This is the upper 95% confidence limit of cancer risk from continuous lifetime exposure to
19 1 /ig/m3 paniculate matter. There is considerable uncertainty regarding this estimate,
20 primarily because of possible species differences in sensitivity and possible variations in
21 slope of the dose-response curve at low exposure concentrations.
22 An RfC of 5 ^g/m3 is recommended. This is a concentration considered unlikely to
23 induce noncancer toxic effects even following continuous lifetime exposure. For the majority
24 of the population there is considerable confidence in this RfC because of general agreement
25 in thresholds from several different animal studies and because of limited responses in
26 occupationally exposed humans. For individuals with existing lung particle burdens near the
27 threshold for inhibition of clearance or induction of toxic effects, however, the RfC may not
28 be protective.
29 Human exposures are estimated to vary from about 1 ^tg/m3 in rural areas up to about
30 15 /ig/m3 in heavily travelled urban streets. Although these are considered reasonable
December 1994 12-22 DRAFT-DO NOT QUOTE OR CITE
-------
1 estimates of mean exposure levels, periodic exposure to much higher, uncertain levels may
2 occur in some portion of the population from time to time.
3 Except for situations such as dwellers near heavily travelled city street canyons,
4 long-term exposures are unlikely to exceed the RfC. The effects of periodic exposure above
5 the RfC, are uncertain although data from occupationally exposed populations suggests any
6 toxic responses are likely to be mild and transient.
7 Based upon the recommended unit risk estimate and an estimate for lifetime exposure of
8 15 jug/m3, the risk to humans could be as high as 5 x 10"4. Because these are upper bounds
9 for both the risk estimate and exposure, the actual risk is likely to be much less for healthy
10 individuals. For individuals either with existing lung particle burdens or exposed to high
11 concentrations of diesel exhaust, risk may be higher. It should be emphasized that these
12 estimates are most valuable for range finding than for estimates of possible disease
13 occurrence.
December 1994 12_23 DRAFT-DO NOT QUOTE OR CITE
-------
1 REFERENCES
2
3 Brightwell, 1; Fouillet, X.; Cassano-Zoppi, A.-L.; Gatz, R.; Duchosal, F. (1986) Neoplastic and functional
4 changes in rodents after chronic inhalation of engine exhaust emissions. In: Ishinishi, N.; Koizumi, A.:
5 McClellan, R. O.; Stober, W., eds. Carcinogenic and mutagenic effects of diesel engine exhaust.
6 proceedings of the international satellite symposium on lexicological effects of emissions from diesel
7 engines; July; Tsukuba Science City, Japan. Amsterdam, Holland: Elsevier Science Publishers B. V.;
8 pp. 471-485. (Developments in toxicology and environmental science: v. 13).
9
10 Federal Register. (1986) Guidelines for carcinogen risk assessment. F. R. (September 24) 51: 33992-34003.
11
12 Garshick, E.; Schenker, M. B.; Munoz, A.; Segal, M.; Smith, T. J.; Woskie, S. R.; Hammond, S. K.; Speizer,
13 F. E. (1987) A case-control study of lung cancer and diesel exhaust exposure in railroad workers. Am.
14 Rev. Respir. Dis. 135: 1242-1248.
15
16 Garshick, E.; Schenker, M. B.; Munoz, A.; Segal, M.; Smith, T. J.; Woskie, S. R.; Hammond, S. K.; Speizer,
17 F. E. (1988) A retrospective cohort study of lung cancer and diesel exhaust exposure in railroad
18 workers. Am. Rev. Respir. Dis. 137: 820-825.
19
20 Ishinishi, N.; Kuwabara, N.; Nagase, S.; Suzuki, T.; Ishiwata, S.; Kohno, T. (1986) Long-term inhalation
21 studies on effects of exhaust from heavy and light duty diesel engines on F344 rats. In: Ishinishi, N.;
22 Koizumi, A.; McClellan, R. O.; Stober, W., eds. Carcinogenic and mutagenic effects of diesel engine
23 exhaust: proceedings of the international satellite symposium on toxicological effects of emissions from
24 diesel engines; July; Tsukuba Science City, Japan. Amsterdam, Holland: Elsevier Science Publishers
25 B. V.; pp. 329-348. (Developments in toxicology and environmental science: v. 13).
26
27 Mauderly, J. L.; Jones, R. K.; Griffith, W. C; Henderson, R. F.; McClellan, R. O. (1987) Diesel exhaust is a
28 pulmonary carcinogen in rats exposed chronically by inhalation. Fundam. Appl, Toxicol. 9: 208-221.
29
30 McClellan, R. O. (1986) Health effects of diesel exhaust: a case study in risk assessment. Am. tnd. Hyg.
31 Assoc. J. 47: 1-13.
32
33 Sienicki, E. J.; Mago, R. S. (1992) Re-evaluation of diesel paniculate emission inventories. In: Toxic air
34 pollutants from mobile sources: proceedings of an international specialty conference. Pittsburgh, PA:
35 Air and Waste Management Association; pp. 151-164.
36
37 Tong, H. Y.; Karasek, F. W. (1984) Quantitation of polycyclic aromatic hydrocarbons in diesel exhaust
38 particulate matter by high-performance liquid chromatography fractionation and high-resolution gas
39 chromatography. Anal. Chem. 56: 2129-2134.
40
41 U.S. Environmental Protection Agency. (1983) Diesel particulate study. Ann Arbor, MI: Office of Mobile
42 Sources.
43
44 U.S. Environmental Protection Agency. (1991a) National air pollutant emission estimates 1940—1990.
45 Research Triangle Park, NC: Office of Air Quality Planning and Standards; EPA report no.
46 EPA/450/4-91/028. Available from: NT1S, Springfield, VA; PB92-152859/XAB.
47
48 U.S. Environmental Protection Agency. (1991b) National air quality and emissions trends report, 1990.
49 Research Triangle Park, NC: Office of Air Quality Planning and Standards; EPA report no.
50 EPA-450/4-91-023. Available from: NTIS, Springfield, VA; PB92-141555/XAB.
51
52 U.S. Environmental Protection Agency. (1993) Motor vehicle-related air toxics study. Ann Arbor, MI: Office of
53 Mobile Sources; EPA report no. EPA/420/R-93/005. Available from: NTIS, Springfield, VA;
54 PB93-182590/XAB.
December 1994 12-24 DRAFT-DO NOT QUOTE OR CITE
-------
1 Woskie, S. R.; Smith, T. J.; Hammond, S. K.; Schenker, M. B.; Garshick, E.; Speizer, F. E. (1988a) Estimation
2 of the diesel exhaust exposures of railroad workers: I. current exposures. Am. J. Ind. Med. 13: 381-394.
3
4 Woskie, S. R.; Smith, T. J.; Hammond, S. K..; Schenker, M. B.; Garschick, E.; Speizer, F. E. (1988b)
5 Estimation of the diesel exhaust exposures of railroad workers: II. national and historical exposures.
6 Am. J. Ind. Med. 13: 395-404.
7
December 1994 12_25 DRAFT-DO NOT QUOTE OR CITE
-------
APPENDIX A
EXPERIMENTAL PROTOCOL AND COMPOSITION
OF EXPOSURE ATMOSPHERES
December 1994 A-l DRAFT-DO NOT QUOTE OR CITE
-------
APPENDIX A EXPERIMENTAL PROTOCOL AND COMPOSITION OF EXPOSURE
ATMOSPHERES1
Facility fSpotaor
Reference
Engine type
Openiinj mode
Fuel type
Fuel aulfur
Exposure refine
Exposure ooodiiKxu
Panicle cooc. (tni/ni*)
Panicle sue
CO, (*)
CO(ppm)
NO, (ppo)
NO(ppo)
SO, (ppoi
SO/: (M*lnK,
Ozone (ppoi)
Aliphattc aldehyde* (pptr)
Formaldehyde (pom}
Acro»tin (pom)
NH4*
THC(ppoi>
PAH*
Bmuxitomi
Benux e ipvrax
ttraa • lanunont
BffTKxt AurvKMnt
Wuormtflwf
>Vm
^^Q\MJmJVtnt
U.S Environmental Proiecuoa A|ency
B(uuu|CT et al., 1980; Campbell et ai_ 1980, 1981:
Hyde et al, 1985; MoomaB a •!.. 1985: Pepdko et
al., 19806, 1981; Pepelio, 1982t>: Pepdko aad
Petnao. 1983; Pkipper et at, 1983
NMU CN 6-33. 3.24 L. 6 by matt
030*0.04
2017*3.01
2.68*030
11.64*234
2.12 tOS6
•
0.177 *0.043
0.106 *0.02V
0.025 *0.003
-
793*1.42
0.52 *0 04
3330*194
437 *1.19
1939 t3X
5.03 si. 03
•
0338 tO.057
0.251 *C.OS9
0.034 tO.009
11.02*1.04
15.9 n'l eonct
28.6 MC/I eanct
53 J nil extract
TJ&H/tanna (k-t-b)
155.8 «*/| extract
198 Mi/| extract
145.2 Mi/I extract
Laune e> al., 1980; Laune and
Boyea, 1980, 1981
374 L 6 cylinder
Federal abort cyck
No. 2diead
0.15%
8 h/d. 7 dMeek. 16 weeta
Coetrol
0.01
0.05 tO.OO*
1^6 *0.06b
0.03 *O.OO*
0.08 *0.01k
0.46 *0.02*
3.22 tO.08"
Ezhauat
5.97 *0.17*
0.28 *0.01k
19.20 *035b
2J1 *0.10*
11.14*043*
1.82*007*
7J9 *0.11k
A-l
DRAFT - DO NOT QUOTE OR CITE
-------
APPENDIX A EXPERIMENTAL PROTOCOL AND COMPOSITION OF EXPOSURE
ATMOSPHERES'
FacihtyftpoMor
Reference
QvvMnr
PMVMTK
lafeacXLZVCtf)
fluonoibm
ImtaxXU-WCd)
W**
B«n»ix^i »•>«•»
U.S. Eaviroomenul Protection Agency
Bhatnafer el •!.. 1980; Campbell el aL, 1980, 1981;
Hyde et •!.. 1985: Moorman et •!., 1985: Pepdko «
«!., 19BOb. 1981; Pepetko, 1982tr, Pepeiko and
Peinno. 1983. Ptopper et at., 1983
71.6 n/i extract
3-5 H/l enract
10.9 «(/| eonct
14^*«/i extract
21.1 *x/| enract
Laune et »!.. 198O. Laune and
Boyea, 1980, 1981
t vt S.D.
A-2
DRAFT - DO NOT QUOTE OR CITE
-------
APPENDIX A. EXPERIMENTAL PROTOCOL AND COMPOSITION OF
EXPOSURE ATMOSPHERES'
Facihry.Sponsor
Reference
Engine type
Operating mode
Fuel type
Fuel sulfur
Exposure refime
Exposure coodutoot
Panicle cone. (nn/tD*)
Panicle toe
C0:(<^
CO (ppinl
NO, (pom)
NO(ppo)}
SO: (pptn)
SO;2 (iMfm1)
Ozone (ppm'i
Aliphatic aWehvdes (ppm)
Formakiehyoe (pom)
Acroieir (ppm,
NH/
THC fppir.i
PAH*
aenio* i .f»' n«
Nvoevrtne
U.S. Envtroomenul Prtxeawo A|et»cv
Wiener el a!.. 1960
NMao CN6-33. 3.24 L 6 cylinder
California cycle, modified
No. 2 dieael
0.15%
20 h/d. 7 diweek. 4 week*
Control
0.00
0.04
10
007
Oil
00
000
00
000
Exhauw
6.32 *1 Jl
01 -10 »im
0.261 *001
174»2J
2.3 tO 4
59 tO.6
2.1 tOS
057 ,052
00
31 6 *2.3
ExhauM-
uradutaj
6^3 1 1 44
0.25 tO 03
16.7 t40
2.9t0.7
5.0 tl.2
1.9 tO£
0.57 »0 13
<0.01
261 s!6
Pepdko ei al.. 1980*
3.24 L 6 cylinder
California cycle, modified
No. 2dMKl
20 h/d. 7 d/week. 4 week*
Exhauti
6.40 »OJ6b
0 26 tO.006*
14.61 tO*f
ZI3 tO.09*
6.13 tO.186
2.10 t0.21»
0.577 *0.019*
31 M il 25"
1 Af t ft S.D. dOM
b SunOMX error of i
A-3
DRAFT - DO NOT QUOTE OR CITE
-------
APPENDIX A. EXPERIMENTAL PROTOCOL AND COMPOSITION OF EXPOSURE
ATMOSPHERES'
Facility Sponsor
Reference
Engine type
Opentuif mode
Fuel type
Fuel sulfur
Exposure regime
Exposure coodiuooi
Panicle cone. (mg/m3)
Panicle size (jtm)
MMD" (GSD)C
CO, {*)
CO (ppm)
N02 (ppm)
NO (ppm)
NO, (ppm)
SO, (ppm)
SO/2 0***)
0,(%)
Ozone (ppm)
Aliphatic aldehydes
Formaldehyde (ppm)
Acrolein (ppm)
NH/
Hydrocarbons (ppm)
PAHs
BmnXiXwrWK
Sivopmne
U.S. Environmental Protection Agency
Pepelko, 1982a
NHMD, 6 cylinder. 3.24 L
California cycle, modified
No. 2 dieacl
20 h/d. 7 dAveek. 4 weeka
Control
Exhauti
6.40 tO.36
0.247 tO.003
16.9*1.1
2.49 *0.18
5.71 *0.21
2.10 *0.21
577 il9
31.6 »3.8
ExhauM •
imdiated
6.75 *OJ9
0.244 *0.007
16.1 tlJ
176 *0.15
4J3 *0.15
146 *0.21
5«9*19
26.1 *34
Lee et al., 1978. 1980
3.24 L 6 cylinder
California cycle, modified
No. 2 (bad
20 h/d. 9 ««eki
Control
0.040
2.0
0.07
0.11
0.0
ZO
Exhaust
6J2
0.252
15.7
2.19
5.85
2.13
0.57
15.6
Exhauat -
irradiated
6.83
0.255
15.4
2.73
494
1.91
0.57
<0.01
150
' All t are standard errors of weekly means.
" Mass median diameter
c Geometric standard deviation.
A-4
DRAFT - DO NOT QUOTE OR CITE
-------
APPENDIX A. EXPERIMENTAL PROTOCOL AND COMPOSITION OF EXPOSURE
ATMOSPHERES'
Facility Sponsor
Reference
En line rvpe
Operating mode
Fuel type
Fuel tulfur
Exposure rep me
Exposure conditiona
Particle cone (mfm*)
Respirabk particles'
(mt'm3)
Particle size (MOD)
MMDC (GSD)"
CO, (*)
CO (ppm)
N02 (ppm)
NO (ppm)
SO, (ppm)
so;: (wn*)
Aliphat. aldehvdc* (ppmt
Formaldehyde (ppm'*
AcetakJehyde (ppm)
Acrotew (pfn)
NH, (ppm)
NH/ (ppm)
THC(pptn)
PAH ((tt/tn*!
BenioUKwranc
taaot • l«near»c«nr
Smart (ftuonnuxrx
FWDT«Kj«-ir
1>
NitKxul Institute for Occupjtioaal Safety ud Health
Cuinnova et il . 1965; Fedan et tl., 1985; Htboo et •!., 1965; Lewii et «!.. 1966. 1989;
Mentnech et al.. 1964; ViUyathan et al., 1966
Caterpillar 3304. 7 L. 4004
008 *OU
0.02 tO 01
0.0076 tO.0035
00015 ±00035
00030*0003?
0.52 tO.28
41 «19
Exhaust
1.95 tO.15
0.23 (t2.5)e
0.36 (t2.0)f
0.20*006
11.5*31
IS *05
8.7 *3.6
0.81 *OJ8
29.0 *24.9
012 ±006
0.0383 tO.0230
0.0387 tO.0153
00602 *00245
064 *071
0 02' tO.0307
7.5 *12 (coW)
135 ±68
196 *o 9
5 6 »2 3
139.3 *96 1
1234 ±72.2
Co«J dust
4.98*0.82
2.00 *0 41
0.09 *0 05
2.2*0.9
0.06 *0 05
0 08 tO.29
0.01 *0.07
16^*179
0.02 tO 01
0.0074 ±0.0041
0.0009 tO 0025
0.0062 tO 0047
0.57 *OJ2
0.0065 ±0014?
3.2 *2.2
26J *11J
32J*15.1
Frturr' 4- coal diui
3.23 sO.60
2.02 *0.30
0.20*007
10.9 *2.8
1.6*OJ
83 *3.2
061 ±029
42.3 *33.8
0 12 *0.05
0.0374 ±0.0266
0.0377 tO.014
0.0578 ±00205
0.48 ±0 55
0.0165 ±00233
7.4 ±2.0 (cold)
10.2 *6.5
11.2*52
3.6 *Z4
67.5 ±5Z4
60.0*366
* All * are S.D,
* < lion
' Ma*s median diameter
6 Geometnc standard deviation
' Electrical aerosol size analyzer
' Scanning electron microscope.
o«herwwe
A-5
DRAFT - DO NOT QUOTE OR CITE
-------
APPENDIX
Facility /Sponsor
Reference
Engine type
Operating mode
Fuel type
Fuel lulfur
Exposure repine
Exposure condition*
Panicle cone (ing/m*)
RespiraWe panicles*
(mt'm3)
Panicle sac ((on)
MMDC (GSD)d
CO: (%)
CO (ppm)
NO, (ppm)
NO(ppni)
SO, (ppa)
SO/2 (Mg/m*)
Aliphatic aldehydes
Formaldehyde (ppm)
Aceialdetryde (ppm)
AcroJeui (ppm)
NH, (ppm)
NH/ (ppm)
THC (ppm)
PAH (uiTr^
Bcnud *»T«r
Nvnpvranc
A. EXPERIMENTAL PROTOCOL AND COMPOSITION OF EXPOSURE
ATMOSPHERES'
National Institute for Occupational Safety and Health
Green ei al., 1963. Rabovtky et al.. 1986
Caterpillar. 7 L, 4 cylinder. 4
-------
APPENDIX A. EXPERIMENTAL PROTOCOL AND COMPOSITION OF EXPOSURE
ATMOSPHERES'
Facility /Sponsor
Reference
Engine type
Operating mode
Fuel type
Fuel sulfur
Exposure regime
Exposure conditions
Panicle cone. (mg/m3)
Panicle sue (*m)
MMD' (GSD)
C02 (%)
CO (tn&Tn')
NO, (ppm)
NO(ppm)
NO, (mgtn-*)
Sulfur (mg.'m3)
SO, (ppm)
AlipbatK aldehyde*
Formalderjyde (ppm)
Acrolcin (pom)
NH/
THC (ppml
PAHs
B«nio(i>«»r»nr
Nxropwrtne
General Motors Research Lab
Barahart et al., 1961, 1982; Cbaudhan et aL, I960,
1961; Chaudhan and Dutta, 1962; Chen and Vostal,
1961; Dziedzic, 1961; Esketoon et aL, 1961; Penney et
al., 1961: Mistorowsto et al., 1960, 1961; Navarre et al..
1961; Schneider and Felt, 1961; Schrec* et aL, 1960,
1961; Strom, 1964; Vostal et al., 1961; Wallace et al.,
1967; White and Garg. 1961
1978 350D Oklsmobile. 5.7 L, 4
-------
APPENDIX A.
Facility Sponsor
Reference
Engine type
Operating mode
Fuel type
Fuel sulfur
Exposure regime
Exposure conditions
Panicle cone (mg/m3)
Panicle size (IUD)
MMD" (GSD)C
CO, (%)
CO (ppm)
NO, (ppm)
NO (ppm)
SO, (ppm)
SO/2 (Mg/m1)
Aliphatic aldehydes (ppml
Formaldehyde (ppm)
Acrolem (ppm)
Ammonia (ppm)
Hydrocarbons (ppffi I
PAHs
Benxo< sOpwene
Nuropynne
EXPERIMENTAL PROTOCOL AND COMPOSITION OF EXPOSURE
ATMOSPHERES*
Inhalation Toocoiofv Research Imtituie
BK* et al.. 1965; Cbeng el al., 1984; Headenon et «L 1983, 1985, 1988; Maudertv et al.. 1983. 1984
1987a. b. 1968; McOdlan et al.. 1986; Wolf a al., 1987
1980 Oldamobile VS. 5.7 L
Federal Tew Procedure, urban driving cyese
PhJllipt No. 2 dieael
034%
7 b>d, 5 d/week, 130 weeks
Control
0.013 tO.006
0 2005 ±0.0390
1.0 ±0.7
0
0
1.1 *3.0
Z6t06
Exhaust
0.353 ±0071
0.183 ±004 (4.8
±0.28)d
0.262 ±0.06 (4.2
±0.24)'
0.2284 ±0.0371
2.9 ±10
0.05 ±0 09
0 7 ±0.3
1.4 ±1.3
3.8 ±09
ExhauM
3.469 ±0.447
0.184 ±0.02 (53 ±
0.64)d
OJ49 ±0.03 (4J
lO-M)*
0.4355 ±0.0590
16J ±7.1
034 ±022
5.7 ±1J
0.9 ±0.9
8.7 ±5.2
Exhjusi
7.082 ±0.808
0.213 ±0.06 (4.7
±0.94)" 0.234 ±0.06
(4.4 ±0.88)«
0.6643 ±0.1320
29.7 ±12.9
0.68 ±0 48
10.0 ±2.6
0.7 ±0.6
13 4 ±8.3
1 All ± are S.D notes* specified otherwise; dau for panicks throufh 30 DO.; dau for
* Mass median diameter.
' Geometric standard deviation.
4 Lovelace multiple jet unpactor. mast median aerodynamic diameter.
' Impactor/parallel flow diffusion battery, mass median diameter.
from 35tb week through 30 mo.
A-8
DRAFT - DO NOT QUOTE OR CTTE
-------
APPENDIX A. EXPERIMENTAL PROTOCOL AND COMPOSITION OF EXPOSURE
ATMOSPHERES'
Facility/Spofl»or
Reference
Engine type
Operating mode
Fuel type
Fuel aulfur
Exposure regime
Exposure coDdiuoos
Panicle cone, (mj/m3)
Panicle toe (pm)
MMDC (GSD^
CO, (%)
CO (ppm)
NO: (ppm)
NO (ppm)
NO, (ppm)
SO, (ppm)
SO/: («g/mx.
O>(<*)
Ozone (ppb)
Aliphatic aldehydes
Formaldehyde (ppm)
Acrolein (ppm)
Ammonia
Hydrocarbon* (ppm)
HTHC (ppm^
PAH»
&er*z.cxi>pvraw
Niti u^n ent
InhalatKM Toocoiofy Reaearcb Inuiiute
Inhalattoo Toncotojy Roean* InMiiuie
- Annual Report, 1960
1980 CM. 5.7 L
California 7-mode urban cycle
PhiUipa No. 2 dM*d
7 h/d, 5 d/weet, 12 weeti
CooiroJ
0039
tO.020
I.I
tO.6
Z8
±0?
Exhaust
0.230
*0.0->3
1 J tO.6
3.2 tO.S
Exhauat
1.030
t0340
3?
tl.l
2.9
t09
Exhaust
4.260
tl.110
0.2080
tO.04
11.5*2.6
04 »04
0.90
»025
146 *31
2J*0.7
4.0 tO^
Mauderfy et »!..
1981*
1980 CM. 57 L
California 7-mode urban cycle
Phillips No. 2 diead
7 htt, 5 d/wcefc. 19week»
Cootro*
0.050
tOO24
Exhauai
0.210
t0070
Eshauti
1.020
»OJ50
ExhauM
4J80
«1 160
* All i are S.D uueu specified oiberwuc.
ConoeniratKxu of caseou* components reported to be proportional to these in 12-
c Mass median diameter.
* Geometric standard deviation
>iudv
DRAFT-DO NOT QUOTE OR CITE
A-9
DRAFT - DO NOT QUOTE OR CITE
-------
APPENDIX A.
F«cility/Spoo»or
Reference
Engine type
Operating mode
Fuel type
Fuel uilfur
Exposure retime
Exposure condition!
Panicle cone (mg/m*)
Panicle size (jun)
MMDb (OSD)C
CO, (%)
CO(ppm)
NOjdJpm)
NO(ppin)
NO, (ppm)
S02 (ppm)
SO;2 (M*/m*)
0;(»)
Aliphatic aldehyde*
Formaldehyde (ppm)
AcroJein (ppm)
NH/
LTHC (ppm)
HTHC(ppm)
PAHi
B«a^nMi^BM
non^yrm
EXPERIMENTAL PROTOCOL AND COMPOSITION OF EXPOSURE
ATMOSPHERES*
Japan Automobile Research Inatitute lac (Health Effect! Reanrch Program • HERP)
HERP. 1988; lahiniahi el al., 1986; UhiaahJ a aL, 19§9
Heavy duty. 11 U 6-cyUnder, direO injection
1200 rpm, eddy current dynamometer
Nippon Oil Co JIS No 1 or 2
0.41%
16 h/d, 6 d/weefc. 30 mo.
Control
0.004
0.068
0.06
0.024
0.040
0.062
0.03
OJS
20.8
0.003
3.62
zx
Exhauu. filtered
0.005
0.083
2J4
042
5.16
5.58
0.9t
1.43
207
0.04
4.43
374
BOratt
OJ9
0.084
2JO
0.44
537
5.81
0.98
57.7
20.7
0.04
4.41
4J3
Ezhauu. filtered
0.019
0.391
13.00
3.96
3Z81
36,76
4JO
1.61
204
0.24
7.79
1Z68
ExhaiMt
Z99
031-0.35
(158-Z83)
0412
12.90
4.95
31 JO
36.45
4.03
358
203
0.20
7.68
13.79
* All * are S.D unJeai tpeafied otnerwwe.
b Maai median diameter.
' Ceometnc tundard deviation.
A-10
DRAFT - DO NOT QUOTE OR CITE
-------
APPENDIX A
Faciliryftponior
Reference
Engine type
Openiinf mode
Fuel type
Fuel »ulfur
Expoaure refine
Exposure ooadiuoai
Panicle cone (mj/m*)
Panicle tize (^m) MMD*
CO, (*)
CO(pptn)
N02 (ppm)
NO (ppm)
NO, (ppm)
SO, (ppm)
SO/2 (Mt/m5)
O, (voJ%)
Aliphatic aldehyde*
Formaldehyde (ppm)
Acroteic (ppm)
NH4*
THC(ppoj)
CH< (ppm)
PAHs (Hfjf nan.):
Bmo*m»
B«n>a(e)f9raK
Bcmdaocfaran*
PtuonntMnc
^ff VK
BowXi AumUWK
BcnuWbXtuonoUMne
Be«l oiherwitc.
b Mast median diameter
A-ll
DRAFT - DO NOT QUOTE OR CITE
-------
APPENDIX A EXPERIMENTAL PROTOCOL AND COMPOSITION OF EXPOSURE
ATMOSPHERES*
Facility /Sponior
Reference
Enpne type
Opentmf mode
Fuel rvpe
Fuel sulfur
Exposure regime
Exposure oonditiom
Panicle cone, (mi/rn*)
Panicle size (*m)b
C02 (%)
CO(ppm)
NOj (ppm)
NO (ppm)
NO, (ppm)
SO (ppm)
SO/2 (Htm*)
O, (vol^t)
Aliphatic aldehydes
Formaldehyde (ppm)
Acrotem (ppm)
NH/
THC(ppm-)
CH4(ppm)
PAH*
NuufMnMC
Fraunbofer Initnui fur Tcoakolope und AeroMtfancbuaf
Heumcfc et a]., 1979; Meiat el ai. 1961
Z4 L
CoiMUnt k»d of 16 kW. 2400 rpm
European reference fuel
OJ6%
74 hAL 5 dMvek. 5 mo.
Control
0.1
<1
<1
6
Exhiusi
4
0.1
OJ
11
0.6
25
26
3
8
i
Exhaau,
filtered
OJ
11
OJ
22
23
4
8
5
Exhauat
11
0.1
0.9
25
U
43
45
8
11
5
Exhavau,
filtered
0.95
27
U
43
44
8
12
5
Exh*uM
17
0.1
1.4
42
2.6
75
78
13
13
5
Exhiuti,
filtered
1.6
45
^•
6K
71
12
13
5
1
* Values esumaied from graphically depicted data.
b Aerodynamic diameter of the modal peak of the panicle man distribution
A-12
DRAFT - DO NOT QUOTE OR CITE
-------
APPENDIX A EXPERIMENTAL PROTOCOL AND COMPOSITION OF EXPOSURE
ATMOSPHERES'
Facility ^Spooaor
Reference
Eafiae type
Operating mode
Fuel type
Fue) tulfur
Expoaurc refine
Expcxurt condition*
Panicle cooc (mi/m>)
Panicle toe (JUD)
JC02(%)
C0(ppin)
NOj(ppm)
N0(ppo»)
NO, (ppui)
SO,(ppni)
SO/2 (Mim**
0:(%)
Aliphatic aldehyde*
Formaldehyde (ppa)
Acroteui (ppa^
NH/
Hydrocarbon* (ppto)
PAHi
Bcmcxtvwvn*
Nwopvranr
Sout)r«t»i Raearcb Inttiiute
KapUn et al., 1983, White el at, 1983
5.7 L
Steady uate, 1347 rpm, equivalent to cooauat 40 mpfa
EmiMioaa 2D
0.23-0^4%
» Ui, 7 d/neek. « weefe
Control
001 *0.009
0.0649 ±0.0020
5.81 tO.2
0
0.05 tO.O
3 43 tO 2
Exhautt
0.242 tO 049
»-93% < 1.0
79*5% < 0.5
0.0781
tO.0028
639*03
036
065*0.1
3.76 t03
Ezhauat
0.735 tO.084
88-94% <1.0
76-84%
-------
APPENDIX A EXPERIMENTAL PROTOCOL AND COMPOSITION OF EXPOSURE
ATMOSPHERES'
Facility /Sponsor
Reference
Enjine type
Opentinf node
Fuel type
Fuel sulfur
Exposure regime
Exposure conditions
Panicle cooc. (m(/m*)
Panicle Mze (jun)
MMD" (OSD)e
CO, (<*)
CO (ppm)
NO, (pptn)
NO (ppm)
NO, (ppm)
SO, (ppm)
SO/2 (MR/ID*)
Oj(
-------
APPENDIX A. EXPERIMENTAL PROTOCOL AND COMPOSITION OF EXPOSURE
ATMOSPHERES'
Facility /Spooaor
Reference
Enpne type
Operating mode
Fuel type
Fuel wlfur
Exposure regime
Exposure cooditkwf
Panick cone, (mf/m3)
Reap piniclei (mi/in*)
Particle tize (JOB)
MMD* (GSD)e
CO, (%)
CO(ppni)
NO, (ppm)
NO(ppm)
NO, (ppcn)
SO, (ppcn)
SO/1 (MI/TO*)
0,(%)
AJiohjuc tkJe+rydci ippcn)
Formaldehyde (ppen)
Acmtan (,ppco>
Amrooou (ppcc)
Hvdrocartjom (poos i
P.\Hi
Bcmot » ipvrw
*•"*—
Batlebe. Pacific Norlhweu Laboniorv
(Carafuno et al., 1961
43 Mtp, 3 cylinder
Sifflulaied mminf cyde
E^wvaJeoi to W-F-800 A gnde DF-2
.
6 h/d, 5 dM>eek. 87 weeka
Control
-
EifcauM
SJtlO
95% reapinbie
0.71 (Z3)
50 «3
44
<1
<1
2^-40
Echauw -f coal dux
13.5 t4.0
' AJI t are S.D. unleu speofied otherwite.
* VU*§ caedun diameier.
c GeootetDc
A-15
DRAFT - DO NOT QUOTE OR CITE
-------
APPENDIX A. EXPERIMENTAL PROTOCOL AND COMPOSITION OF EXPOSURE
ATMOSPHERES'
Facility /Sponsor
Reference
Engine type
Operating mode
Fuel type
Fuel sulfur
Exposure refime
Exposure condiDon*
Panicle cone, (mj/m*)
Panicle size (USD)
C02 (%)
CO (ppm)
NO, (ppo)
NO (ppm)
NO, (pptn)
S02 (ppns)
SO/2 ((ti/m*)
0, (%)
Aliphatic aldehydes
Formaldehyde (ppm)
Aceialdehvde
Acrolein (ppm)
NH4*
Hydrocarbon* (ppm)
Benzene (ppm)
Toluene (ppm)
PAHi (Mita3):
Bnoiopvrinc
Ncropvnne
Univenity of Pittsburgh
BMiicdli. 1965
7 bp. four cycie, tuigte cylinder
15-60 mm
Dilution A
0.1
<»
1J
0.2
20.5
<1.0
<01
<0.05
<10
Dilution B
0.9
30
2.8
0^
200
-------
APPENDIX A. EXPERIMENTAL PROTOCOL AND COMPOSITION OF EXPOSURE
ATMOSPHERES*
Fadbry,5ponaor
Reference
Enpoe type
Openimf node
Fuel type
Fud wlfur
Exposure regime
Exposure coodiuow
Particte oooc.
(mj/m3)
Panicle size (JUD)
CO, (<*)
CO(ppa)
N0j(ppn>)
N0(ppm)
NO, (ppm)
SOjCppo)
HjSO, (ppm)
OxxUnu
(ppm a* C3)
Aliphatic aktefavda
Fonaakteliyoe
(ppos>
Acrotetn (pperO
NH<*
Hydrocsrtxxa
(ppw at CH^I
PAH«
B«io(i)pwtnr
Nurop^tne
U.S. EDviroomeoul Protedioo Agency
Gilletpie, 1980: Hyde et «!.. I960; VUlaachuk. I960; Onboefer, I960, Sun et •!.. I960
Automobile fMoiioe engine
Urt«a cyde
16 hM. 7 tKweet 68 mo.
Control
4.9
0.04
0.04
0.03
o.o:
17
Nc*-irr»di*ied
(Moiiae
oduutt (R)
97 J tlO.O
0.05 tO.02
1.45*042
-
-
-
27.5 i4 4
Imdiaied
|Mo4ioe
eduusi (D
94 J t!9.6
0.94 tO.36
0.19 tO.29
•
-
020*0.09
23.9 t6.1
SO,*
K,S04
•
•
•
0.42
tO.22
002
tOOl
-
-
R *
S0:*
H^04
96.4
t!3£
0.05
tO.03
Ul
tO.44
048
*023
0.02
tOOl
-
27.4
*4J
1 +
SO,*
HjSO,
-
OJ9
tOJ6
0.19
tO.29
0.42
*0.21
0.03
tOOl
0.20
to.oe
23.9
t6.0
Nitrofen
oode»
-
0.64
tO. 12
075
*0.06
•
-
-
-
Nitrofce
ondea
-
0.15
t033
1.67
*0.21
-
-
-
-
1 Ai; t are S D. untest specified oiherwiae.
A-17
DRAFT - DO NOT QUOTE OR CITE
-------
APPENDIX A.
Facility /Sponsor
Reference
Engine type
Operating mode
Fuel type
Fuel sulfur
Exposure repine
Exposure
conditions
Panick cone
(m^ta*)
Panick we (JUB)
MMD' (GSD)C
CO. (*)
CO (pom)
NO, (pom)
NO (pom)
NO, (pom)
SO, (pprn)
SO/2 (Mi/m3)
Oj(%)
Aliphatic
aldehvdes
FormaJbeiiyoe
(ppm>
Acroteio (pperl
NH(*
Hydrocarbon* (pom)
HTHC (.ppm)
PAHs (ns/m*)
B«mo<.*ivr*ne
Banoftlflu rnataa
B«u>(atiiXMnfe»K
I-Niiranmc
EXPERIMENTAL PROTOCOL AND COMPOSITION OF EXPOSURE ATMOSPHERES*
Japan Automobik Research Institute IDC (Health Effects Research Profram • HER?)
HER? 1968: lahiiwhi et al . 1986: laniniaoi et ai, 1989
Lifbt duty, 1^ L. 4
-------
APPENDIX B
CONTRACTOR REPORT:
ASSESSMENT OF RISK FROM EXPOSURE TO
DIESEL ENGINE EMISSIONS
December 1994 B-l DRAFT-DO NOT QUOTE OR CITE
-------
SUMMARY
Significant numbers of diescl locomotives began to appear on U.S. railroads in the 1940s,
and by 1959, virtually all U.S. locomotives were powered by diesel engines. This report
describes an effort to conduct a quantitative assessment of the risk of lung cancer from
exposure to diescl exhaust (DE) based upon data from a retrospective cohort study of
lung cancer mortality among U.S. railroad workers (Garshick et a/., 1988) and an
industrial hygiene study of exposures of U.S. railroad workers to DE (Woskie et al.,
1988a,b; Hammond et al., 1988).
Garshick et al. (1988) studied information obtained from the Railroad Retirement
Board (RRB) on 55,407 white males who began railroad employment between 1939 and
1949, who were between the ages of 40 and 64 in 1959, and who in 1959 worked at one
of 39 jobs selected to represent a range of potential exposures to DE. Garshick et al.
(1988) reported two analyses that indicated an effect of exposure to DE upon lung
cancer risk in this cohort:
1) A relative risk for lung cancer of 1.45 (95% CI = 1.11, 1.89) was observed
for DE-exposed workers who were 40-44 years of age in 1945 and who
consequently had the longest potential exposure to DE; relative risks were
progressively lower among DE-exposed workers who were older in 1959 and
who thus had potentially shorter exposures to DE.
2) The relative risk of lung cancer increased monotonically with increasing
duration of work in 1959 or later in a job involving diesel exposure
(disregarding exposures in the current year and in the four most recent
years); this relative risk was 1.72 (95% CI * 1.29, 2.33) in the group with the
longest exposure (15-17 years).
Concentrations of respirable paniculate matter (RSP) were collected in four small,
northern U.S. railroads during 1981-1983 (Woskie et a/., 1988a; Hammond et al., 1988).
B-l DRAFT-DO NOT QUOTE OR CITE
-------
This information was collected for 13 jobs categories, which were obtained by grouping
the 39 jobs used to define the Garshick et at. (1988) cohort. In addition to RSP, two
additional markers of DE were developed: adjusted respirable paniculate (ARP), which
is determined by subtracting from the concentration of RSP an estimate of the
concentration of environmental tobacco smoke (ETS), and adjusted extractable matter
(AEM), which is the concentration of RSP that is extractable by dichloromethane minus
an estimate of the concentration of the extractable fraction of ETS present. Significant
differences were found between concentrations of ARP in samples collected on warm
days (>10°C) and cold days (<10°C). Because of this temperature effect, the
concentration of ARP was estimated separately for warm and cold days for each job
group.
In the present study, data from Woskie et al (1988a) and Hammond et al (1988)
were used to assign average exposures to each cohort member by year, beginning in 1959
based on yearly job codes. In addition to RSP, ARP and AEM, exposure to total
extractable material (TEX) was also studied. TEX is the concentration of extractable
RSP, not adjusted to remove the contribution of ETS. Since ETS is also suspected of
being a risk factor for lung cancer, it is plausible that TEX might be more closely
correlated with lung cancer than measures that omit ETS.
Temperature-specific estimates were developed for each of these four markers,
and a temperature-weighted estimate of each worker's yearly exposure to each marker
was obtained, based on the location of the last railroad where he was employed and
temperature data for that location. As a hedge against the possibility that the
temperature-specific estimates were biased for some reason, ARP, unadjusted for
temperature, was also studied.
Job codes were not available for years prior to 1959, so it was assumed that during
those years of employment a person worked in the same job as in 1959. Exposures in
each job category were assumed to begin after 1945 and to increase linearly by year
through 1959, when dieselization of U.S. railroads was virtually complete.
Poisson regression using internal controls was the principal statistical method used
to study the relationship between measures of diesel exposure and lung cancer mortality.
Measures of exposure considered were: cumulative exposure omitting the most recent
B-2 DRAFT-DO fiOT QUOTE OR CITE
-------
three years; cumulative exposure 4-8 years in the past; cumulative exposure 9-18 years in
the past; cumulative exposure more than 18 years in the past, and maximum exposure in
any year prior to the three most recent years. Diesel exposure was also dichotomized as
yes/no, with clerks and signalmen considered to be unexposed.
Measures of cumulative exposure during different time intervals in the past were
studied in order to investigate different potential latency periods for lung cancer induced
by DE, i.e., to determine whether exposures during a particular period in the past had a
greater influence on lung cancer mortality than exposures during different periods. If this
was the case, then including exposures during other time periods could mask an effect of
diesel exposure. Measures based on maximum exposure were considered to explore an
#•
hypothesis suggested by animal data that clearance of respirable panicles from the lung is
capacity limited and, as a result, more intense exposures may be more dangerous per
amount inhaled than less intense exposures.
Other explanatory variables considered in the analyses were age, calendar year
and job category (clerk; signalman; engineer and firer; brakeman, conductor and hostler;
shop worker). Both relative risk models and absolute risk models were applied to the
data.
More than 50 analyses of the relationship between exposure to DE and lung
cancer mortality were conducted. These involved five markers of exposure to DE, five
ways of accumulating previous exposures, severaJ subgroups of the cohort based on job in
1959, and both relative risk and absolute risk models. None of these analyses
demonstrated a pattern that was consistent with an adverse effect of diesel upon lung
cancer; in fact, many of them showed a statistically significant negative association.
Consequently, we were not able to project a dose response for lung cancer induced by
DE or estimate the carcinogenic potency of DE from these data.
The data available in Woskie et al. (1988a) and Hammond et al. (1988) to
estimate exposures to DE in the cohort studied by Garshick et al. (1988), and the
methods applied to these data, have a number of limitations, including:
• Use of markers of exposure that are not specific for DE;
• Lack of personal sampling data for members of the cohort;
B-3 DRAFT-DO NOT QUOTE OR CITE
-------
• Extrapolating exposures measured in four small railroads in a limited
geographical region to the entire U.S.;
• Assuming exposures measured during 1981-1983 were representative of earlier
exposures;
• Lack of knowledge of the extent to which individual workers worked with
diesel equipment during the period of changeover from steam to diesel;
• Adjusting exposures for temperature differences based on limited information
on the area served by the last railroad at which a worker was employed.
Because of these limitations, there is considerable uncertainty regarding the
estimates of exposure of the cohort to DE. These uncertainties are potentially of
sufficient magnitude to obscure any relationship between exposure to DE and lung
cancer that may exist in the cohort. Consequently, these negative findings may be due to
weaknesses in the exposure data rather than to an absence of diesel effect, per se.
Analyses in which diesel exposure was considered as a yes/no variable confirmed
the first finding of Garshick el al. (1988) Further analyses in which workers in specific
diesel-exposed jobs were compared to unexposed workers revealed that:
The risk of lung cancer among engineers and firers was significantly
elevated relative to that among unexposed railroad workers, and the
variation in relative risk with age in 1959 (increasing risk with
decreasing age) is consistent with DE being responsible for the
»
observed excess.
Although the risk of lung cancer among conductors, brakemen and
hostlers was significantly elevated relative to that among unexposed
railroad workers, the variation in risk with age in 1959 (higher at older
and younger ages than at intermediate ages) is not consistent with an
effect of DE.
The risk of lung cancer among shop workers was not significantly
elevated relative to that of unexposed railroad workers. This is not
B-4 DRAFT-DO NOT QUOTE OR CITE
-------
consistent with an effect of DE upon lung cancer because it seems
likely that shop workers had significantly higher exposures to DE than
either of the other two groups of exposed workers.
An age-specific plot of percentage of deaths in this cohort by calendar year
reveals that age-specific death rates for all causes remain fairly constant at values roughly
comparable to corresponding rates in U.S. white males until about 1977. Beginning in
about that year, death rates begin to decrease year-by-year, and in 1980 they are from
two-fold to six-fold smaller than corresponding age-specific rates in U.S. white males.
The only plausible explanation for this that we can envision is that a significant number
of unrecorded deaths must have occurred in the cohort after 1977. A new tape has
recently been provided by the RRB that includes follow-up of the cohort for additional
years, as well as an update of the follow-up through 1980 (Garshick, 1991). It is reported
that for 1980, about 25% of the deaths on the updated tape were not on the earlier tape.
Some additional deaths in 1979 are also on the new tape, but the percentage is much
smaller. However, our work suggests that the lack of follow-up may have been more
severe than this.
A Poisson regression analysis of cumulative years of work in 1959 or later in a job
involving diesel exposure (disregarding exposures in the current year and four most
recent years) found an inverse relationship between cumulative years of exposure and
relative risk of lung cancer. The relative risk of lung cancer for 1-4 cumulative years of
exposure was 1.38 (95% CI = 1.17, 1.62), for 5-9 years of cumulative exposure was 1.28
(95% CI = 1.10, 1.49), for 10-14 years of cumulative exposure was 1.10 (95% CI = 0.92,
1.3), and for 15-17 years of exposure was 1.06 (95% CI = 0.75, 1.50). Although the
reason for the differences in these results and those of Garshick et al. (1988) has not
been completely determined, it appears to be related to the fact that the three variables
used in the analysis - calendar year, age, and cumulative years of exposure - are all
correlated, which makes the analyses highly sensitive to the specific method used for
controlling for age and calendar year. Whereas Garshick et al. (1988) modelled age as a
continuous variable and used calendar year to define the risk sets in a Cox regression
B-5 DRAFT-DO NOT QUOTE OR CITE
-------
analysis, we controlled for calendar year using an indicator variable for each year and
controlled for age using an indicator variable for each five-year age interval.
Follow-up of railroad workers' mortality in this study was through 1980, which is
only 22 years since the dieselization of U.S. railroads was essentially complete. Since the
time from first exposure until evidence of an increased risk of environmentally induced
lung cancer is often on the order of 20 years, the full impact of any effect of DE upon
lung cancer in this cohort may not be captured by the current study. Considering this,
and also considering the fact that follow-up may have been incomplete in the latter years
of the study, it would be worthwhile to conduct a new study of this cohort to take
advantage of the several additional years of follow-up now available. If such a study is
*r
conducted, it is recommended that vital status be verified independently of RRB records.
INTRODUCTION
Diesel engines emit a much larger quantity of respirable particles than gasoline
engines of comparable size (McQellan, 1987). Associated with these panicles are
carcinogenic and mutagenic compounds, which suggests that diesel exhaust (DE) may be
carcinogenic when inhaled. A considerable amount of research, including
epidemiological studies and in vitro and whole animal bioassays, have been conducted to
evaluate the carcinogenic potential of DE.
Extracts from DE have been shown to be active (e.g., mutagenic) in a number of
short-term in vitro assays and carcinogenic in SENCAR mouse skin tumor assays (Lewtas
and Williams, 1986). Exposure of rats to DE at high levels for a prolonged period has
resulted in excess lung tumors (Mauderly et ah, 1986).
Howe et al. (1983) reported an elevated risk of lung cancer among railroad
pensioners who were formerly exposed to DE and coal dust. The study did not contain
information on duration or levels of exposure to DE.
Garshick et al (1987) conducted a case control study of lung cancer deaths among
U.S. railroad workers. A significantly increased relative odds (odds ratio = 1.41) was
observed among men who were 64 years old or younger at the time of death, and who
had worked for 20 years or longer in a job involving diesel exposure. No effects of DE
B-6 DRAFT-DO NOT QUOTE OR CITE
-------
were seen in workers 65 or older; however, it was reported that many of these workers
retired shortly after the transition to diesel locomotives was complete.
Garshick et al. (1988) reported on a retrospective cohort study of 55,407 U.S.
railroad workers who began railroad work in 1950 or earlier and were followed from
1959 through 1980. The findings of this study are summarized in the next section. As
part of this study, an industrial hygiene study was conducted between 1981 and 1983 of
exposure of U.S. railroad workers to DE (Woskie et al, 1988a,b; Hammond et al., 1988).
A number of other epidemiological studies of diesel-exposed populations have not
found an increase in cancer (Kaplan, 1995; Waxweiler et a/., 1973; Waller, 1980; Hall and
Wynder, 1984; Wong et a/., 1985).
The present study describes an effort to make quantitative estimates of the cancer
risk posed by exposure to DE using epidemiological data from the Garshick et al. (1988)
retrospective cohort study of U.S. railroad workers and information on the exposures of
these wcrkers to DE from Woskie et al (1988a,b) and Hammond et al (1988).
DESCRIPTION OF EPIDEMIOLOGICAL DATA
Review of Study of U.S. Railroad Workers
Using Interstate Commerce Commission (ICC) job codes Garshick et al (1988)
identified 39 job groups for study. Job groups selected involved large groups of workers,
with and without regular exposure to DE. Major categories of jobs assumed to involve
substantial exposure to DE were shop worker, engineer, firer, brakeman, conductor, and
hostler. Clerical workers (clerks, station agents, dispatchers, etc.) and signalmen were
considered to be unexposed.
The RRB provided data on workers in these job groups based on job codes for
1959. For their analysis, Garshick et al (1988) selected a cohort consisting of 55,407
white males who began their employment between 1939 and 1949 and were between the
ages of 40 and 64 in 1959.
The follow-up period for the study was from 1959 through 1980. Workers not
reported by the RRB to have died by December 31, 1980, were considered to be alive.
Cause-specific death certificates were obtained for 88.3% of the deaths; remaining deaths
were assumed to be due to unknown causes.
B-7 DRAFT-DO NOT QUOTE OR CITE
-------
Partial likelihood methods (Cox, 1972), using calendar year to define risk sets,
were used to compare the lung cancer mortality in diesel-exposed and non-diesel-exposed
workers. Age was controlled for by estimating category variables for five categories of
age in 1959. Also calculated was the relative risk of lung cancer by categories of years of
exposure to DE, compared to non-exposure, using age as a continuous variable.
The age-specific relative risk of lung cancer in exposed workers versus unexposed
workers was 1.45 (95% CI * 1.11-1.89) for those aged 40-45 and decreasing in an almost
monotonic fashion to a relative risk of 0.99 among workers aged 60-64 in 1959. Similar
analyses of relative risks among workers potentially occupationalry exposed to asbestos
versus workers not so exposed did not show a relationship with potential asbestos
exposure. The results pertaining to diesel exposure were reproduced by independently
calculating directly standardized relative risks for lung cancer among diesel workers
versus unexposed workers for each five-year age group in 1959.
Analyses of relative risk versus years of exposure did not find a consistent
exposure-response relationship. However, if exposure in the year of death and the four
most current years were omitted, the group with >,15 years of exposure had a relative
risk of 1.72 (95% CI = 1.29, 2.33). Smaller but statistically significant relative risks were
obtained for 1-4, 5-9, and 10-14 years of cumulative exposure. These relative risks
increased monotonicaUy with increasing years of exposure.
The higher relative risks among persons who were younger in 1959 are consistent
with an effect of DE upon lung cancer because the younger groups would have had a
longer time remaining to work and be exposed to DE. Similarly, higher relative risks
among persons with longer exposures to diesel are consistent with an effect of DE upon
lung cancer.
Epktemiologic Data Available for This Assessment
Data on U.S. railroad workers used in the Garshick et al. (1988) study were
provided to us on tape by Dr. Garshick and Dr. Speizer. Table 1 contains a catalog of
the data received. ICC_1 through ICC_22 are the yearly job codes (ICC, 1951) provided
to the RRB. A list of the sampled job codes for 1959 used to define the cohort and the
number of workers in each category are shown in the top portion of Table 2. Some
B-8 DRAFT-DO NOT QUOTE OR CITE
-------
workers were assigned job codes in years subsequent to 1959 that were not among those
sampled. These unsampled job codes appear in the bottom half of Table 2.
There were 19,384 deaths in the cohort from all causes. Deaths were considered
due to lung cancer if the death certificate listed ICC code 162.0 or 1611 as either the
primary or secondary cause of death. There were a total of 1,640 deaths from lung
cancer as determined by the primary cause of death and 54 (3%) as determined by the
secondary cause of death, for a total of 1,964 deaths from lung cancer.
DESCRIPTION OF EXPOSURE DATA
Data used to estimate exposure to DE in this cohort come from two published
papers concerning measurements of markers for diesel exposure collected from four
small railroads in the northern U.S. between 1981 and 1983 (Woskie et al., 1988a;
Hammond et al., 1988). Measuring exposure to DE is complicated by the fact that diesel
paniculate is a complex mixture containing respirable carbonaceous particles upon which
many organic chemicals may be adsorbed. Because of the complex chemical nature of
diesel paniculate, it is very difficult to distinguish diesel paniculate from respirable
paniculate matter from other sources.
To deal with this problem, Woskie et al. (1988a) and Hammond et al. (1988)
considered several markers for diesel paniculate in addition to the concentration of total
respirable paniculate. Woskie et al. proposed adjusted respirable paniculate (ARP)
concentration as a marker for diesel exposure. ARP is estimated by subtracting from the
concentration of respirable paniculate matter (RSP) an estimate of the concentration of
environmental tobacco smoke (ETS). In addition, Hammond et al. proposed the
concentration of adjusted extractable matter (AEM) as a marker for DE. AEM is
defined as the concentration of RSP that is extractable by dichloromethane minus an
estimate of the extractable fraction of ETS. Railroad workers can be exposed to
inorganic respirable particles from a variety of sources, including welding material,
fibrous material and aerosolized sand (Hammond et a/., 1988). Use of AEM as a marker
for diesel paniculate can help eliminate interference from respirable inorganic paniculate
matter.
B-9 DRAFT-DO NOT QUOTE OR CITE
-------
The 39 job codes sampled by Garshick €t al. (1988) were combined into 13
homogeneous job groups and over 550 samples of RSP were collected in these 13 job
groups. The job groups were selected so that exposures within a group represented
essentially the same type of work and proximity to running locomotives. Each sample
was collected over a single work shift. Personal breathing zone samples were collected
from all job categories except clerk and engineer. Area samples were collected for these
latter jobs, as workers in these jobs were considered to remain essentially stationary.
Each sample was categorized according to outdoor temperature as warm (mean outdoor
temperature > 10°C) or cold (mean temperature < 10°C).
B-10 DRAFT-DO NOT QUOTE OR CITE
-------
Table 1
Contents of Data Tape on Mortality of U.S. Railroad Workers Used in Analysis'
Variable Name
IDENT
AGE59
ASB59
BANO
COD1
COD2
DEATHL12
FIRST_ES
FIRWRK
ICC 1 to ICC 22
EXPOS 1 to EXPOS 22
MODC
MON 1 to MON 22
MSURV59
RACE
RET
SRVC
SSN
YODH
YOBH
Description
Unique identification number
Age in 1959, 40-64
PotentiaJ for asbestos exposure based on job code in 1959,
defined as HI or NO
Last railroad employer
Underlying cause of death from the death certificate based
on ICDA-8
"Secondary" cause of death, coded only if cancer or chronic
lung disease was noted on the death certificate and it was
not the underlying cause of death
l=lung cancer death, COD1 or COD2, 0= death not due to
lung cancer
Estimated year first worked for the railroad, based on total
service months and year last worked for the railroad
Year first worked for the railroad, provided by the Railroad
Retirement Board, but exists only for 1947 or greater
Yearly 3 digit job codes, 1959-1980
Yearly job category. 1959-1980
Month of death
Months of railroad service by year, 1959-1980
Survival in months from January 1959 until death or end of
the follow-up period
All white males; RACE-1
Retirement year (year last worked for the railroad)
onths of total railroad service
Social security number (dead cases only)
Year of death
Year of birth
"Provided by Dr. Garshick.
B-11 DRAFT-DO NOT QUOTE OR CITE
-------
Table 2
Job Codes for Cohort
Garshick et al. (1988) Exposure Group
Job Category
ICC Coda
Number of Men
A. SAMPLED JOB CODES IN 1959 (Used to define cohort)
Clerk
Signalman
Passenger Engineer/Fireman
Yard Engineer/Fireman
Freight Engineer/Fireman
Passenger Brakeman/Conductor
Freight Brakeman
Freight Conductor
Yard Brakeman/Conductor
Hostler
Machinist
Electrician
Supervisor (Shop)
a UNSAMPLED JOB
007,012,075,076,078-
083
044-049
121,125
124,128
122,123,126,127
111.112,116
117,118
113,114
119,120
108-110
061
058
050
10,475
3.548
881
4,446
5,676
1,148
6,134
1,089
9.126
780
6.635
4,288
1.169
CODES
Clerk
Signalman
Shop Workers
Brakeman
Other Train Riders
Carman
Unknown Job
No Job/After Retirement
001-006,008-01 1,013-016,019-022,026,077,
084-087,099,104-106,129,301-311,317-318
017-018,024,028,035-043,073,074,088-094,
098,102-103312-316319-325
052-055,062-066,068,070-071
107
095-097,100-101,115 (for 115, EXPOS
assigned as 6)
023,025,027,029-034,051,056, 057,059,060,
067.069,072390-395,200.202-204,21 1
000
no ICC Code present in array
B-12 DRAFT-DO NOT QUOTE OR CITE
-------
A cyclone was used to insure collection of respirable particles only. The amount
of respirable paniculate on each filter was determined as the difference in the filter
weight prior to and subsequent to collection of the sample.
After determining the amount of RSP from each filter, the filters from three of
the railroads were composited by job group and railroad and extracted with
dichloromethane. One-half of the extracted material was analyzed by gas chromatograph
to determine the amount of nicotine present. The concentration of ETS in each pooled
sample was estimated by comparing the amount of nicotine in each extract to the amount
of nicotine in experimentaJ samples of RSP known to be composed solely of ETS.
Information on ETS was not available for one of the four railroads or for some of
the job categories in the remaining three railroads. The average estimated concentration
within a job group was used to estimate the average concentration for the missing data in
that group. The concentration of ARP was estimated for each job group across all
railroads by subtracting the applicable average fraction that was due to ETS from each
railroad's average RSP concentration. A weighted average concentration for each job
category was calculated using the number of samples from each railroad as the weights.
The half of extracted material not used for nicotine determination was combined
across the three railroads by job category and weighed to determine the amount of
extractable material present. The concentration of AEM was then calculated for each
job category as the concentration of RSP times the fraction extractable minus the
estimated concentration of ETS times the fraction of ETS extractable.
Table VII of Woskie et al. (1988a) contains average concentrations of AEM for
each of 13 job groups, categorized by "Cold" and "Warm". For some jobs there are
significant differences between concentrations for "Cold" and "Warm" days. Exposures in
the indoor jobs of clerks and shop workers are about twice as high on cold days as on
warm days, which is consistent with improved ventilation on warm days. On the other
hand, other jobs involve higher exposures on warm days.
Development of Measures of Diesel Exposure for LLS. Railroad Workers
Development of Temperature-Specific Estimates of Exposure. The differences in
average concentrations of AEM on cold and warm days obtained by Woskie et al.,
suggest that estimates of individual exposures could be improved by taking temperature
B-13 DRAFT-DO NOT QUOTE OR CITE
-------
differences into account, based upon the geographical area in which a cohort member
worked. The first step in accomplishing this was to develop temperature-specific
estimates of exposure for exposure measures other than ARP. Table II of Hammond
et al. contains estimates of average concentrations of RSP and AEM obtained from three
railroads. These estimates were made temperature-specific by multiplying a
concentration of RSP or AEM from Hammond et oi's Table II by the ratio of the
temperature-specific concentration of ARP from Woskie et a/.'s Table VII to the
corresponding (temperature independent) estimate for ARP from Hammond et o/.'s
Table II.
In addition, a fourth marker of exposure was developed called total extractable
material (TEX). Temperature-specific estimates of TEX were developed by multiplying
the percent extractable mass from Hammond et a/.'s Table III by the corresponding RSP
concentration in Hammond et a/.'s Table II and then making the estimates temperature-
specific as described in the previous paragraph. RSP contains material other than diesel
paniculate that may be carcinogenic to the lung (e.g., tobacco smoke), so it is of interest
to determine the extent to which TEX is a predictor of lung cancer incidence.
The resulting temperature- and job-specific exposure estimates are tabulated in
Table 3. Estimating the temperature-specific values for the other markers from
temperature-specific values for ARP is a relatively crude approach, because temperature
gradients in exposures may be different for different exposure markers. However, this
approach provides a reasonable approximation and, unfortunately, temperature-specific
data are not available for the other exposure markers of interest.
In order to use the temperature-specific estimates of exposure, temperature
variables were developed for each member of the cohort to reflect the prevailing
•+
temperature patterns in the area in which an individual worked. The only railroad-
specific information available for members of the cohort was a code indicating the name
of the last railroad for which the person had worked. It was assumed that the
temperature patterns in the location served by this railroad would be representative of
his entire work history.
Each railroad has designated one or more individuals as contact officials who
interacted with the RRB on various matters. Upon request, the RRB furnished the
B-14 DRAFT-DO NOT QUOTE OR CITE
-------
names and address of the contact person for each railroad. These addresses were used
to assign a geographical area (state) to each railroad. In some cases the area of
operation was ambiguous and The Official Guide of the Railways, May, 1970" was used
to designate a state of operation of a railroad. According to information obtained from
the RRB, there are a limited number of railroads that operate in multiple geographic
regions (Class I railroads) and the operation of most other railroads is confined to the
immediate area surrounding their headquarters (Ferguson, 1990).
A publication of the National Oceanic and Atmospheric Administration (NOAA,
1978) contains climatological information for various U.S. cities by month, including the
average temperature, average maximum daily temperature and average percent of days
that were clear, cloudy or partly cloudy, based on data from 1941 through 1970. Data
from a centrally located city in each state were selected as representative of
climatological data for that state.
B-15 DRAFT-DO NOT QUOTE OR CITE
-------
Table 3
Climate Adjusted Exposures (jig/mj)
Job Description
Clerk
Signalman
Passenger Engineer/
Fireman
Yard Engineer/Fireman
Freight Engineer/
Fireman
Passenger Brakcman/
Conductor
Freight Brakeman
Freight Conductor
Yard Brakeman/Conductor
Hostler
Machinist
Electrician
Supervisor (Shop)
ARP
42
58
51
69
94
104
102
69
114
224
147
194
155
ARP
Cold
54
33
18
65
108
79
110
80
150
276
181
257
160
Warm
25
74
68
72
67
123
94
52
56
172
95
113
40
AEM
Cold
9.3
13.1
8.5
15.1
34.5
20.5
8.5
34.8
22.4
40.7
44.0
79.0
44.4
Warm
4.3
29.3
32.0
16.7
21.4
31.9
7.3
22.6
8.4
25.3
30.0
32.0
11.1
RSP
Cold
167.1
40.4
43.1
69.7
114.9
85.1
138.0
148.4
252.6
287.1
231.0
254.0
250.8
Warm
77.4
90.6
162.7
77.2
71.3
132.5
118.0
96.5
94.3
178.9
127.0
118.0
62.7
TEX
Cold
81.9
16.6
17.7
28.6
47.1
23.8
31.7
69.8
78.3
45.9
73.9
109.2
110.4
Warm
37.9
37.1
66.7
31.7
29.2
37.1
27.1
45.3
29.2
28.6
53.3
50.7
27.6
NOTE: Cold corresponds to < 10°C.
Warm corresponds to >10°C.
B-16 DRAFT DO NOT QUOTE OR CITE
-------
Since most workers were considered to work primarily during daylight hours, the
average daytime temperature (average temperature between 8:00 A.M. and 5:00 P.M.)
was estimated for each month and each state. Numerical integration was used to
calculate the average 24 hour temperature (AT), the average daytime temperature
(ADT), and the maximum temperature (MT), based on a typical daily temperature cycle
(Figure 1). These three quantities were used to calculate a temperature factor (TF)
defined by the following equation:
ADT = AT + TF * (MT - AT).
Separate temperature factors were estimated for clear and cloudy days:
TF* = 0.77 (clear)
TF,^ = 0.51 (cloudy)
These temperature factors were then used to estimate the average daytime
temperature for each month from the NOAA (1978) data as follows:
{AVERAGE DAYTIME MONTHLY TEMP]-
[AVE. MONTHLY TEMP.] * ([AVE. MAX. DAILY TEMP. FOR MONTH]
- [AVE MONTHLY TEMP.])) •
[FRACTION CLEAR DAYS] + [FRACTION PARTLY CLOUDY DAYS] +
— —
[FRACTION CLOUDY DAYS] * [FRACTION PARTLY CLOUDY DAYS]
-
,
Average temperatures for the years 1941 through 1970 were used in these calculations.
If [AVERAGE DAYTIME TEMP] was above 10°C for a particular month, that month
was designated as a warm month; otherwise it was designated as a cold month. These
calculations were performed for each state. A summary of these calculations is provided
in Table 4.
To calculate a worker's "climate-adjusted'1 exposure to a particular marker of
diesel paniculate for a given year, a weighted average of the Varm" and "cold" exposures
(listed in Table 3) corresponding to his job category for that year was calculated, with the
weights being the proportion of warm or cold months listed in Table 4. These weighted
exposures were assumed to apply from 1959 onward.
B-17 DRAFT-DO NOT QUOTE OR CITE
-------
O*arday
28 —
Figure 1
26*
V
•s 22'
<2
Hour* el tw
IS 18
21 2«
Th« dMv t«no*r«ttf« CVCM on a dMr vanui a ctoudv otv
Clevn cow wi( o«of»»4 A* o*iy mawnum
mvwnunt tametraur*
Sourct: Oliver (1987).
B-18 DRAFT-DO NOT QUOTE OR CITE
-------
No address was available for a number of the smaller railroads. Since no climate-
adjusted exposures could be developed for workers in these railroads, analyses involving
climate-adjusted exposures omitted these workers. There were 897 workers in this
category (1.6% of cohort).
Assigning Exposures in Unsampled Job Categories. The 13 worker categories in Table 3
that were assigned exposures correspond to the 13 categories of job codes for 1959 that
were sampled in the Garshick el al (1988) study, and which are listed in Table 2. The
lower part of Table 2 contains a list of job codes that were assigned in subsequent years,
but were not sampled in 1959. An unsampled job was assigned the same exposure as a
sampled job that appeared to involve similar exposures. "Other train riders" were
assigned to the same exposure category as conductors, brakemen, and hostlers, and
"carmen" were assigned to the same category as clerks. Persons with an unknown job in
a given year (ICC code 000) were assigned to the same exposure category as the previous
year. These assignments should have a very limited effect because the unsampled job
codes corresponded to only a small fraction of the total person-years of follow-up (other
train riders, 0.2%; carmen, 0.5%; unknown job, 0.5%).
Estimation of Exposures Prior to 19S9. Since job codes are not available prior to 1959,
workers were assumed to have had the same job from the beginning of employment to
1959 as they had in 1959. Since information is not available concerning which workers
worked near diesel equipment during the years prior to 1959 in which both diesel and
steam equipment were used, it was assumed that exposures in each job category
increased linearly with time from 1945 through 1959. This linear increase parallels the
overall rate at which the locomotive fleet became dieselized (AAR, 1989).
The first year of employment by a railroad was specified in the data base provided
by the RRB for only 2,908 workers (2%), all of which began work in 1947 or later.
However, the total months of railroad service was available for each worker, and the
starting date of work was estimated based on total service months and last year of work.
B-19 DRAFT-DO NOT QUOTE OR CITE
-------
Table 4
Number of Months per Year the Average
Temperature is Greater than 10°C by State
STATE
Alabama
Alaska
Arizona
Arkansas
California
Colorado
Connecticut
Delaware
Florida
Georgia
Hawaii
Idaho
Illinois
Indiana
Iowa
Kansas
Kentucky
Louisiana
Maine
Maryland
Massachusetts
Michigan
Minnesota
Mississippi
Missouri
MONTHS
12
5
7
9
12
7
6
8
12
12
12
7
7
7
6
9
8
12
6
8
6
6
6
12
8
STATE
Montana
Nebraska
Nevada
New Hampshire
New Jersey
New Mexico
New York
North Carolina
North Dakota
Ohio
Oklahoma
Oregon
Pennsylvania
Rhode Island
South Carolina
South Dakota
Tennessee
Texas
Utah
Vermont
Virginia
Washington
West Virginia
Wisconsin
Wyoming
MONTHS
6
6
6
6
8
9
6
9
6
7
9
8
7
6
12
6
9
12
7
6
9
7
9
6
6
B-20 DRAFT-DO NOT QUOTE OR CITE
-------
This approach will overestimate the start date for workers whose employment was not
continuous. To give some idea of the error introduced by this approximation, the
estimated first year of work was compared to the actual year for the 2,908 workers whose
starting year was known. In this sample, the estimated starting year was correct for 2,336
workers (80%), one year too high for 463 workers (20%), two years too high for 104
workers (3.6%) and three years too high for five workers (0.2%). Thus, in this sample, at
least, the method of estimating the starting year did not introduce gross errors. It should
be noted that this method will provide an accurate estimate of total years of exposure
prior to 1959, although some of the exposures may have occurred somewhat earlier than
estimated.
METHODS OF ANALYSIS
Methods Involving Poisson Regression
The principal analytic method used to evaluate the relationship between exposure
variables and lung cancer incidence was Poisson regression. [See BEIR IV (1988) and
BEIR V (1990) and references contained therein for a detailed discussion of the
method.] In this approach, each explanatory variable (e.g., age, calendar year, exposure
measure, etc.) is categorized by dividing its range into subintervals. This categorization is
used to cross-tabulate the data into cells, with each cell containing data corresponding to
a particular category (subinterval) of each explanatory variable. Each cell will thus
contain data corresponding to similar values of each explanatory variable. Data
tabulated into cells include the person-years of follow-up and the number of cases. The
average value of one or more of the explanatory variables may also be calculated for
each cell, with the average being weighted by the number of person-years corresponding
to the cell. Since the values of some explanatory variables for a single individual in the
cohort may vary with time (e.g., cumulative exposure), the experience of a single cohort
member may not be summarized in a single cell but rather contribute to a number of
cells.
The number of cases occurring in each cell is assumed to have a Poisson
distribution with the expected number of cases being the product of the number of
person-years in that cell and the risk per person-year predicted by a statistical model.
The data are then fit by the process of maximum likelihood (Cox and Undley, 1974) and
B-21 DRAFT-DO NOT QUOTE OR CITE
-------
hypothesis tests and confidence intervals are computed based on a large sample
maximum likelihood theory.
Several explanatory variables were selected that were considered to be potentially
important predictors of lung cancer risk in this cohort. Non-exposure variables used in
the analysis included age, calendar year, and job category in 1959.
In most of the analyses, age was divided into five 10-year intervals: 40-49, 50-59,
60-69, 70-79, and 80+. In some of the analyses, age was divided into ten 5-year intervals
beginning with 40-45. A few of the analyses were conducted twice using both divisions of
age in order to evaluate the effect of subdividing age in different ways.
Calendar year was categorized into either seven intervals (1959-1965, 1966-1968,
1969-1971, 1972-1973, 1974-1977, 1978-1979, and 1980) or 22 intervals (each interval
corresponding to one calendar year). Job code in 1959 was categorized mainly according
to five major divisions: clerk; signalman; engineer and firer, conductor, brakeman, and
hostler; and shop worker. However, exposures were estimated separately for each of the
13 job groups listed in Tables 2 and 3.
Four different exposure variables were created for each exposure measure. These
were: cumulative exposure omitting most recent three years (current year and two
previous years); cumulative exposure in past four through eight years (omitting most
recent three years); cumulative exposure in past nine through 18 years; and cumulative
exposure more than 18 years in the past. The first of these exposure variables was
categorized into six subintervals and the remaining three generally into four subintervals.
Gass boundaries were set so that roughly equal numbers of person-years were in each
subinterval. These different exposure variables were used in order to investigate
different potential latencies for lung cancer induced by DE; i.e., to determine whether
exposures during a particular period in the past were more highly correlated with lung
cancer risk than exposures during other periods. If this were the case, then including
exposures during other periods could mask an effect of diesel exposure.
These exposure variables were created for each of the five markers of diesel
exposure considered (ARP, climate-adjusted ARP, climate-adjusted AEM, climate-
adjusted RSP, and climate-adjusted TEX).
B-22 DRAFT-DO NOT QUOTE OR CITE
-------
There is some evidence from animal studies that the clearance of respirable
particles from the lung is capacity limited (Morrow, 1988). This would suggest that more
intense exposures may be more dangerous per amount inhaled than less intense
exposures. To explore this hypothesis, a maximum exposure variable was created for
each of the five markers of diesel exposure, defined as the maximum one year exposure
occurring more than three years in the past. If more intense exposures are the most
dangerous, then lung cancer risk could correlate more closely with maximum exposure
than cumulative exposure.
Several types of statistical models were applied to the cross-tabulated data. First,
the risk of lung cancer was modelled as
y, eb
where risic^ is the risk per person-year in the cell corresponding to the ith age category,
jth calendar year category, and the kth exposure category, and s^, yf and ek are estimated
parameters. The dose response was indicated by plotting the exposure parameters ek,
along with 90% confidence intervals, versus the average exposure in the kth exposure
subinterval.
Second, the risk of lung cancer was modeled as
risk^e) = a, yt [1 + ft ej,
where riski/e) is the risk in the ith age group and jth calendar year category from an
exposure e. The exposure used in applying the model was the average exposure in a cell.
In this model a single dose parameter, 0, is estimated rather than a different parameter
for each dose category. The latter approach gives a more complete description of the
dose response, whereas the former approach summarizes the dose response in a single
variable, ft. A positive value of ft corresponds to a positive relation between exposure
and lung cancer and a negative value for ft corresponds to an inverse relationship (higher
exposures being associated with lower risks).
B-23 DRAFT-EX) NOT QUOTE OR CITE
-------
Both of the above models are multiplicative models, i.e., they assume that the
effect of diesel exposure is to multiply the background risk of lung cancer. A few
analyses were based on the additive model
a, y{ + a'^.
In this model diesel exposure is assumed to increase the background risk of lung
cancer in an age-dependent additive manner.
In addition to being applied to the complete cohort, these models were also
applied to various subsets of the cohort defined by job in 1959, or by age in 1959.
In some of the analyses, exposure was a dichotomous variable (yes/no). As in
Garshick et al (1988), clerks and signalmen comprised the unexposed group in these
analyses.
Methods Using Comparisons with US. Rates
One analysis was conducted to compare the cancer and overall death rates in this
cohort with comparable rates from the general U.S. population. Observed numbers of
lung cancer deaths in various categories were compared with expected numbers of deaths
and based on age- and calendar year-specific mortality rates for U.S. white males.
Methods Using Partial Likelihoods
An analysis based on a proportional hazard model implemented using the partial
likelihood method of Cox (1972) was used to model the dependence of relative risk of
lung cancer upon elapsed time from beginning of exposure (1959) to five years prior to
death or retirement. In contrast to the application used by Garshick et al. (1988), the
risk sets were defined by age rather than by calendar year. Twenty-two category
variables, one for each year, were used to control for calendar year. Workers classified
as clerks or signalmen in 1959 were considered to be unexposed (i.e., these workers were
assigned an elapsed time of zero).
The software package EPICURE (Preston et al, 1990) was used to implement all
of the analyses in this report.
B-24 DRAFT-DO NOT QUOTE OR CITE
-------
RESULTS
Comparisons of Ripened to Unoposcd Worken Independent of Exposure Level
Table 5 contains the observed and expected number of cancers (expected
calculated from age- and calendar year-specific death rates for U.S. white males), and the
associated relative risks, by age in 1959 and by exposed and unexposed workers. The last
column of Table 5 contains the ratios of relative risks for exposed workers to those for
unexposed workers. This column confirms the finding by Garshick et al (1988) of higher
risks in exposed workers relative to unexposed, particularly among the workers who were
younger in 1959. The ratios of relative risks in Table 5 are almost identical to the
relative risks appearing in Table 5.
Table 5
Lung Cancer Deaths in Garshick et al. (1988)
Age in
1959
40-44
45-49
50-54
55-59
60-64
Unexposed Worken
Observed
Deaths
67
69
90
92
64
Relative
Risk
0.63
0.69
0.87
1.03
1.12
Exposed Worken
Observed
Deaths
291
309
280
285
147
Relative
Risk
0.90
0.93
0.97
1.22
1.11
RR Exposed/
RR Unexposed
1.43
1.34
1.12
1.18
0.99
As discussed by Garshick et aL (1988), since the younger workers in 1959 were the
ones with the potentially longest exposures, the trend indicated by the last column of
Table 5 is consistent with an effect of diesel exposure upon lung cancer. However, Table
5 reveals some features of the lung cancer rates that were not apparent from the internal
analysis conducted by Garshick et aL (1988). Within each group of workers (unexposed
and exposed), the risk relative to U.S. rates increases with age in 1959. However, this
increase is greater among unexposed than exposed workers, which accounts for a
decreasing relative risk with age in 1959 in internal comparisons of exposed and
unexposed. Consequently, using U.S. rates as a comparison population, relative risk
increases, rather than decreases, with age in both the exposed and unexposed groups.
B-25 DRAFT-DO NOT QUOTE OR CITE
-------
In order to explore the reason for this behavior, the pattern of both lung cancer
deaths and total deaths by calendar year were studied. Figure 2 displays graphs of the
percent of persons, by year, alive at the beginning of the year and who die of any cause
during the year. The figure contains a separate graph for each five-year age group
beginning with ages 40-44. The graphs for most age groups do not extend across all
years because there were no cohort members in some age groups in certain years. (E.g.,
since everyone in the cohort was at least 40 yean old in 1959, by 1964 there were no
longer any cohort members aged 40-44.)
Up through about 1976, the age-specific death rates remained roughly constant,
which is consistent with the pattern of death rates for the U.S. during this period.
However, beginning in about 1976 the death rates begin to decrease precipitously with
time. By 1980 the death rates in men aged 80-85 was similar to that of men aged 60-65
between 1959 and 1976. Similarly, in 1980, the death rates for men aged 65-70 were
similar to those aged 50-55 between 1959 and 1976. Although there were too few deaths
from lung cancer to reliably graph such trends for each five-year age interval, the total
death rate for lung cancer decreased after 1976 in a manner that paralleled the decrease
in total deaths.
Figure 2 also indicates the death rates for U.S. white males for the years 1959 and
1980. Although the rates for 1980 are somewhat lower than those for 1959, rates for
these two years are similar and both sets of rates are comparable to the death rates in
the cohort between 1959 and about 1976. However, death rates in the cohort begin to
fall below U.S. rates in years subsequent to 1976. By 1980, the rates for the cohort were
below U.S. rates by factors ranging from two (for ages 60-65) to almost six (for ages
greater than 85). Thus, not only did death rates in the cohort decrease after 1976, they
t-
were comparable to U.S. rates prior to 1976 and decreased to values considerably below
U.S. rates in subsequent years.
Consideration was given to the reason for the observed drop-off in mortality in
this cohort. It is highly unlikely that it is an effect of diesel exposure. The possibility that
it was a manifestation of the health worker effect was also considered. However, to the
extent that such an effect is associated with cancer response, it should have been more
apparent in the early years of follow-up rather than in later years.
B-26 DRAFT-DO NOT QUOTE OR CITE
-------
JS
*
1
u
12 -
10 -
8 -
i
TfrT,
•Mt-
JO-W-
Figure 2
Percent Deaths by Year and by Age
-S
M^BA
•9*0
•u-ti
JO-W
I I I
i s i i i i i
40-44
65-69
45-49
70-74
50 -54
75-79
55-59
80-S4
60-64
85
B-27 DRAFT-DO NOT QUOTE OR CITE
-------
The only plausible explanation that has been proposed for the drop-off in
mortality is that it is due to a lack of follow-up in the cohort; deaths must have occurred
that were not included in the file provided by the RRB.1 The RRB was the only source
of death ascertainment in the Garshick et al. (1988) study, so there was no verification of
death ascertainment through an independent source.
Since there is evidence that not all deaths were recorded from about 1977 onward,
analyses were conducted to determine if the patterns seen in Table 5 still appear if
follow-up was stopped at the end of 1976. Table 6 shows the results of a Poisson
Table 6
Relative Risks of Lung Cancer Death in Exposed
Versus Unexposed Workers for Different Follow-up Periods
Age
in 1959
40-44
45-49
50-54
55-59
60-64
FoDow-up through 1980
Exposed vs. Unexposed
(95% CI)
1.49 (1.1 - 1.9)
1.31 (1.0 - 1.7)
1.12 (0.88 - 1.4)
1.17 (0.92 - 1.5)
0.96 (0.74 - 1.3)
Follow-up through 1976
Exposed vs. Unexposed
(95% CI)
1.64 (1.2 - 2.3)
1.25 (0.93 - 1.7)
1.20 (0.93 - 1.6)
1.12 (0.88 - 1.4)
1.17 (0.70 - 1.3)
regression analysis based on the complete data set, and secondly, based upon a truncated
data set in which follow-up is assumed to end on December 31, 1976. When follow-up
continues through 1980, the results of the Poisson regression agree closely with both
1 It has recently been reported (Garshick, 1991) that a new tape provided by the
RRB includes follow-up of the cohort for additional years, as well as an update of the
follow-up through 1980. For 1980, about 25% of the deaths on the updated tape were
not on the earlier tape. Some additional deaths are also on the new tape for 1979, but
the percentage is much smaller. However, Figure 2 suggests that the lack of follow-up
was more severe than this.
B-28 DRAFT-DO NOT QUOTE OR CITE
-------
those in Table 5 obtained by comparison with U.S. rates, and those reported in Garshick
et al. (1988) Table 5 obtained using partial likelihood methods. When follow-up is
stopped at the end of 1976, the results do not change materially. There is still a trend
toward higher risks in exposed workers relative to unexposed workers who were younger
in 1959, and the excess lung cancer among exposed workers is statistically significant in
men aged 40-44 in 1959. Moreover, the relative risk is higher among workers aged 40-44
in 1959 when follow-up stops in 1976 than when it continues through 1980. This finding
indicates that the excess relative risk among exposed workers found by Garshick el al.
(1988) holds even if data after 1976 are omitted from the analysis.
Table 7 indicates the relative risk in exposed versus nonexposed for different
groups of exposed workers. Among engineers and firers, the relative risk is significantly
elevated among those who were 40-44 in 1959 and nearly so among those who were
45-49 in 1959. The data for engineers and firers also exhibit the monotone relationship
between relative risk and age in 1959 that is seen in the complete cohort of exposed
workers (Table 6). This group also exhibits an almost significant deficit of lung cancer
relative to unexposed workers among those 60-64 years of age in 1959.
Table 7
Relative Risks of Lung Cancer in Exposed
Versus Unexposed for Different Exposed Groups
Age
in 1959
40-44
45-49
50-54
55-59
60-64
Engineers and
Firers
1.71 (1.2, 2.4)
1.34 (0.97, 1.8)
1.44 (1.1, 1.9)
1.04 (0.76, 1.43)
0.70(0.45, 1.1)
Brakemen, Conductors
and Hostlers
1.58 (1.2, 11)
1.29 (0.97, 1.7)
1.08 (0.81, 1.4)
1.28 (0.97, 1.7)
1.26 (0.88, 1.8)
Shop Workers
1.13 (0.79, 1.6)
1.30 (0.94, 1.8)
0.92 (0.68, 1.2)
1.15 (0.87, 1.5)
0.92 (0.64, 1.4)
Brakemen, conductors and hostlers exhibit a relative risk greater than unity in
each age category, and the excess is statistically significant among workers 40-44 years of
B-29 DRAFT-DO NOT QUOTE OR CITE
-------
age in 1959. However, the pattern of higher relative risks among workers who were
younger in 1959 is not present; rather, relative risks are higher in the youngest and oldest
age categories and lower in the middle age categories.
Among shop workers, the relative risk is not significantly elevated in any age
group. Furthermore, there is no statistically significant evidence of an increased risk
among shop workers, either overall (p * 0.13 based on a one-sided test of significance of
a single relative risk applied across all age groups) or in any five-year age group based on
age in 1959. There is also no discernable pattern of variation of relative risk with age in
1959.
Analyses of Cumulative Yean of Exposure or Elapsed Yean Since Beginning of
Exposure
The partial likelihood analysis of relative risk with elapsed time from 1959 until
four years prior to death or retirement found a statistically significant excess relative risk
among exposed workers in the two lowest categories of elapsed time but not in the two
highest, and the excess did not increase monotonicalty with elapsed time. The relative
risk of lung cancer with 1-4 years of elapsed time was 1.25 (95% CI = 1.04, 1.51), for 5-9
elapsed years was 1.34 (95% CI « 1.15, 1.56), for 10-13 elapsed years was 1.15 (95% CI
= 0.97, 1.36), and for 15-17 years was 1.33 (95% CI * 0.98, 1.82). A corresponding
analysis based on Poisson regression gave almost identical results. The relative risk of
lung cancer with 1-4 years of elapsed time was 1.26 (95% CI = 1.04, 1.52), for 5-9 years
was 1.34 (95% CI - 1.14, 1.56), for 10-13 elapsed years was 1.14 (95% CI - 0.96, 1.35),
and for 15-17 elapsed years was 1.32 (95% CI • 0.97, 1.79). In this analysis, calendar
year was controlled for using an indicator variable for each year, and age was controlled
for using an indicator variable for each five-year age interval.
A Poisson regression analysis found an inverse relationship between cumulative
years of exposure lagged five years and excess relative risk. The relative risk of lung
cancer for 1-4 cumulative years of exposure was 1.38 (95% CI = 1.17, 1.62), for 5-9 years
of cumulative exposure was 1.28 (95% CI » 1.10, 1.49), for 10-14 years was 1.10 (95%
CI = 0.92, 1.3), and for 15-17 yean was 1.06 (95% CI * 0.75, 1.50). In this analysis,
persons classified as clerks, signalmen or carmen during a year were considered to be
unexposed during that year. A person with a missing job code was assigned the job code
for the previous year. Cumulative years of exposure were calculated by summing the
B-30 DRAFT-DO NOT QUOTE OR CITE
-------
months of exposure by year and dividing by twelve. Calendar year was controlled for
using an indicator variable for each year and age was controlled for by using an indicator
variable for each five-year age interval. Exposures during the current year and the most
recent four years were ignored.
Dose Response Analyses
Table 8 summarizes the results of 50 dose response analyses in which exposure
was treated as a continuous variable. An estimated dose coefficient and associated
standard error is provided for each analysis. A positive dose coefficient indicates a
positive relationship between exposure and lung cancer risk, whereas a negative
coefficient indicates an inverse relationship. Although the model being applied assumes
a linear dose response, a positive dose response should result in a positive coefficient
even if the dose response is non-linear.
All but two of the 50 analyses produced a negative dose coefficient.2 The two
exceptions involved signalmen, who are considered to have had only limited exposure.
Many of the negative values were statistically significant (absolute value of the dose
coefficient greater than twice the standard error).
Figures 3 through 12 summarize the results of selected dose response analyses in
which independent dose coefficients are estimated for each exposure interval. In each
figure, the risk in the lowest exposure group has been normalized at 1, and the risk in
each exposure group relative to that in the lowest exposure group is plotted at the
average exposure in that group, along with the 90% confidence interval on the relative
risk.
2A number of these analyses produced negative coefficients so large that negative
probabilities were estimated for a few cells. To allow the program to converge, a few
cells were deleted from the analysis whenever this occurred. The deleted cells contained,
at most, only 2.4% of the total person-years included in an analysis.
B-31 DRAFT-DO NOT QUOTE OR CITE
-------
Figure 3
Lung Cancer Risk Versus Unadjusted ARP
K>
8
O
H
O
c
J
• I
«*
12
*4
o
jn
n
-------
8
O
H
O
c
o
o
50
O
a
Figure 4
Lung Cancer Versus Unadjusted ARP
Excluding Signalmen. Shop Workers, and Hostlers
i-
0«
01
ot
-------
Q.
(T
•o
0)
to
0)
"03
LO g
c/)
g> c/5
U.
O
c
03
O
O)
c
1
B-34 DRAFT-DO NOT QUOTE OR CITE
-------
Q_
CC
T3
Q)
1o
co
E
OJ
g>
LL
C/3
CC
o
c
co
a>
1
I
CO
x
!•
y«M (•> I
12
Do**
iM
B-35 DRAFT-DO NOT QUOTE OR CITE
-------
Figure 7
Lung Cancer Risk Versus Climate Adjusted AEM
O
C
O
a
o
po
O
11
• t
i
o/
• 4
—I
MM
—I
MO
-------
Figure 8
Lung Cancer Risk Versus Climate Adjusted AEM
V
O
1
O
po
0
Excluding Signalmen. Shop Workers, and Hostlers
01
ot
04
01
200
Amiaga OOM to *•*» <• r*"1
-------
Figure 9
Lung Cancer Risk Versus Climate Adjusted RSP
8
2:
O
H
O
c
3
o
50
n
i-
o*
• t
• 3
It >
••••OdM kitten*
It
—I—
It
-------
Figure 10
Lung Cancer Risk Versus Climate Adjusted RSP
u*
O
c
O
a
O
73
O
a
ExcludinQ Signalmen, Shop Workers, and Hostlers
• i
ot
01
•
o»
1 • F
I* >
ao* DOM fc «•*• «»r«"«
—I—
at
—r—
it
-------
Figure 11
Lung Cancer Risk Versus Climate Adjusted TEX
&
D
8
O
H
O
§
n)
o
50
O
00
I
I"
Of
00
ot
04
01
—r~
04
00
00
12
-------
Figure 1
Lung Cancer Risk Versus Climate Adjusted TEX
o
s
"Z.
O
H
O
C
O
O
po
O
a
Exckidlno Signalmen, Shop Workers, and Hosllers
13
12
00
or
o«
o»
04
02
04 00 00
/W«t«0* Dot* n |*j*u •>
-T™
II
-------
Each of these figures shows the dose response for cumulative exposure four or
more years in the past corresponding to one of the five markers for diesel exposure.
Figures are presented for both the entire cohort and for the subcohort defined by
omitting data on signalmen, hostlers, and shop workers. These jobs were omitted in
order to emphasize jobs for which exposure estimates are considered to be the
mostreliable. A number of similar plots using other exposure variables are shown in
Appendix E. None of these plots (including those in Appendix E) indicate a positive
relationship between a marker for diesel exposure and risk of lung cancer. Indeed, as a
whole, they tend to indicate a negative relationship.
Since the highest relative risk in Table 5 is for persons aged 40-45 in 1959, the
analyses in Table 8 and Figures 3 through 12 were repeated for cumulative exposure to
unadjusted ARP more than three years in the past for the complete cohort,
engineers/firers, conductor/brakemen, and shop workers, but restricted to persons aged
40-45 in 1959. These analyses did not indicate a relationship between exposure and lung
cancer. The dose response slope was negative for the complete cohort, engineer/firers,
and conductors/brakemen. Although the slope was slightly positive for shop workers, it
was not statistically significantly different from zero.
Table 9 presents results from applying the Poisson regression model in which
cumulative exposure to ARP (unadjusted for climate) more than three years in the past
was entered categorically in an additive fashion. As with the analyses in which exposure
was entered in a multiplicative fashion (Figures 3 through 7), this analysis suggests a
negative relationship between cumulative exposure to DE and lung cancer. Similar
results were obtained when this model was applied to the other four markers of diesel
exposure.
Maximum one-year exposure occurring more than three years in the past was not
positively correlated with lung cancer relative risk for any of the five markers of diesel
exposure considered. In fact, the dose response slope was negative for each of the five
markers.
B-42 DRAFT-DO NOT QUOTE OR CITE
-------
Table 9
Relation Between Lung Cancer Mortality and Exposure
to ARP Using an Additive Risk Model
Average
Exposure to ARP
(Mg/m3-years)
320
650
880
1210
1660
2390
Increase in Lung Cancer
Risk Relative to Lowest
Exposure Group
0.0
•0.18
-0.95
-0.82
-0.94
-1.6
95% Confidence
Interval
_
(-0.44, -0.086)
(-1.3, -0.63)
(-1.1, -0.49)
(-1.3, -0.54)
(-2.4, -0.85)
Comparison of Results of This Study with Other Risk A
nts
Several studies have made quantitative assessments of the potential risk of lung
cancer from diesel exposure based on both bioassay and epidemiological data available.
Harris (1983) estimated a range for the carcinogenic potency of DE based on the
negative epidemiologic study of London Transport Workers (Waller, 1980). Harris also
estimated the potency of DE in humans using a comparative analysis of laboratory and
epidemiological data on DE and two chemically related environmental exposures - coke
oven emissions and roofing tar emissions. Smith and Stayner (1990) estimated the
potency of DE from a rat bioassay of Mauderty et al. (1986). Albert and Chen (1986)
made a similar application of the Mauderty et al. data; they also utilized a comparative
potency approach somewhat similar to that applied by Harris. After reviewing both
epidemiological data as well as short-term bioassay data on exposure to several sources
of potycyclic aromatic hydrocarbons (PAHs), including DE, Cuddihy et al. (1984)
projected an upper limit to the carcinogenic potency of DE.
Table 10 summarizes the carcinogenic potencies obtained from these studies,
expressed in units of increase in relative risk of lung cancer per jjg/m'-year exposure to
B-43 DRAFT-DO NOT QUOTE OR CITE
-------
Table 10
Estimates of the Increase in Relative Risks
of Lung Cancer from Exposure to DE
Harris (1983)
Cuddihy et aJ. (1984)
Albert and Chen (1986)
Smith and Stayner
(1990)
Method
Epidemiological analysis of
London transport workers
Average relative potency
for extracts from three
automobiles
Range from relative
potency of extracts from
Caterpillar engine
Review of epidemiological
and short-term assay data
Multistage model applied
to Mauderh/ et al. (1987)
data on rats
Relative potency
Multistage time-to-tumor
model applied to Mauderly
et al. (1987) data on rats
Increase in Relative Risk
(xlO**) per Mg/m3-year*
MLEb = 1.2
95% upper bound = 4.3
95% lower bound = -1.8
0.35
0.01 - 0.25
<0.46
MLEC = 0.012
95% upper bound = 0.039
0.050
0.046 - 0.092
* Assumes occupational exposure (40 hours/week, 50 weeks/year).
b MLE - Maximum Likelihood Estimate.
e Albert and Chen (1986) only provided 95% upper bound.
DE. Computations required to express the results,, from the various studies in these units
are provided in Appendix E.
Estimates of the carcinogenic potency of DE obtained from these studies vary by
more than two orders of magnitude. The largest estimate is the upper limit of 4.3x10"*
per fig/m'-year obtained by Harris from a negative epidemiological study. The estimates
derived from lung cancer induced in long-term animal inhalation studies range from
0.012X10"4 per /ig/m3-year to 0.092X10"4 per Mg/m3-year. Even though the upper limit
derived by Harris is much larger than the estimates derived from the animal data, the
B-44 DRAFT-DO NOT QUOTE OR CITE
-------
epidemiological data used by Harris are not inconsistent with the animal results because
the estimates based on animal data fall between the lower and upper limits estimated by
Harris.
To obtain some idea of how these estimates compare with the results of this study,
Figure 13 shows a graph of the relative risks measured in the cohort as a function of
ARP. Superimposed upon this graph are plots of the relative risks predicted by the
upper limit and maximum likelihood estimate (MLE) obtained by Harris from
epidemiological data and the largest estimate based upon the animal data. All of these
plots are clearly inconsistent with the relative risks from this study plotted as a function
of cumulative exposure to ARP. The width of the confidence intervals obtained from
this study compared to the excess relative risk predicted by the plots suggests that, due to
statistical variability alone, it would be virtually impossible for a study involving the
numbers of subjects and range of exposures in the Garshick et al (1988) study to detect
an excess risk of the magnitude indicated by the animal studies. However, it appears
that it might be possible to demonstrate a statistically significant relationship between DE
and lung cancer if the effect of DE is as large as that predicted by the upper limit
obtained by Harris from the epidemiologicaJ data.
DISCUSSION
A number of methods of analysis were used to search for a relationship between
markers of exposure to DE and lung cancer mortality. Five different markers of diesel
exposure were investigated, four of which were adjusted for climatic differences.
Measures of exposure studied included cumulative exposure (four or more years in past),
as well as cumulative exposure in various time periods in the past (4-8 years, 9-18 years
and 18 or more years in the past). If lung cancer mortality from DE is closely associated
with exposure in one of these prior time periods, then including exposures in other time
periods could mask an effect of diesel exposure.
B-45 DRAFT-DO NOT QUOTE OR CITE
-------
i ^ — 1
^ i
I
\ ,
1..
I
V i
N i
\
^
V 1
\
\ 1
\
\ '
\ 1
\
! \ 1
£ W «B ^ '
f 5 1 ' N
«» •« - \
3 - ! « '
£ 2 i
_ • a.
^R s 5 i?
o n 1 •
k_ ^5 d > 8
3 °£ 2i
O O CD a k
C £"5 |3
\
\
\
\ 1
\
\ |
\ \
II If '> •
® 5 CL X
OC »T i * i
s? t i x
Sl 1 \ '
"o w B
c s i
s^
« JC
0
'
i
> |
\
•
I L .
'
» t
! * "N ^ ^|
I
«(
\ * !
•, i
i
\'i I
-
_ IA
CM
-
— W • "^~~
!j:
s. i
-: s 1!
•
E
fn '
•
•* t"'
1 i-
• j
•~ ~ c
1
* i
' _•
=
'* '
in
o
"
B-46 DRAFT-DO NOT QUOTE OR CITE
-------
Maximum exposure in any single year more than three years in the past was also
analyzed. Consideration of this exposure variable was motivated in part by animal results
that indicate that clearance of inhaled particles is less efficient at higher doses, and
consequently more intense exposures may be more dangerous per amount inhaled than
less intense exposures. No relationship between maximum exposure and lung cancer was
found. However, the correlation between this measure of maximum exposure and actual
exposures in the population is apt to be poor. Thus, even if more intense exposures
were more dangerous per amount inhaled in this cohort, it would probably be difficult to
demonstrate this fact since individual measures of DE were not available.
Analyses were conducted using both relative risk and absolute risk models. These
methods were applied both to the complete cohort and to subgroups defined by job
category in 1959. Single job groups were selected for analysis, as well as combinations of
job groups that were thought to have the most reliable exposure information.
None of these analyses showed a pattern that was consistent with an effect of
diesel exposure upon lung cancer. Consequently, it was not possible to develop a model
for lung cancer mortality as a function of exposure to DE from these data.
There are several potential reasons for this negative finding: 1) the quality of the
measures of markers of exposure to DE available may be too poor to reveal an
association with lung cancer mortality; 2) confounding with other determinants of lung
cancer risk may be masking the effect of DE; 3) there may be no causal relationship
between lung cancer risk and exposure to DE, or the relationship may be too small to be
detected in a study of this magnitude, given the exposures to diesel experienced by this
cohort. Each of these possible causes will be discussed individually below.
Quality of Exposure Information
Both the exposure data and the epidemiological data are of critical importance in
developing a quantitative dose response model. Deficiencies in either cannot be
counterbalanced by exceptional quality in the other.
The exposure data used in this assessment came from air samples collected during
a limited time period (between 1981 and 1983) at four small railroads operating in a
limited geographical area (northern U.S.). Measured concentrations were of RSP rather
than DE per se, and these measured concentrations were adjusted to produce markers of
B-47 DRAFT-DO NOT QUOTE OR OTE
-------
diesel exposure. These data were used to estimate exposures to DE that occurred
among railroad workers throughout the U.S. as much as 30 or more years in the past.
Dearly, this approach has many potential limitations.
Diesel equipment and working conditions have changed since the 1940s when
diesel engines first began being used in large numbers. Woskie et al. (1988b) described
anecdotal reports of smoky working conditions in the diesel repair shops during the 1950s
and 1960s. They also report that limited data available on nitrogen oxide levels during
these periods qualitatively support higher levels of DE in the early years of dieselization.
By the time samples were collected for this study, these smoky conditions would have
been largely mitigated through improved ventilation and the advent of less smoky
engines.
Woskie et al. report that the types of diesel equipment used by the sampled
railroads during the sampling period were representative of the type and ages of
locomotives used nationally from the 1950s through the early 1970s. Further, the mixture
of diesel equipment used by the four sampled railroads was similar in type and age to
what might be expected nationwide. However, there was no way to verify that this
mixture was actually representative of conditions nationwide during the period (1945-
1980) that the cohort was exposed.
Railroad workers are exposed to emissions from a number of sources that may be
confounded with DE in exposure estimates. These include ETS, aerosolized sand,
fibrous material, and welding fumes (Hammond et a/., 1988). The five markers of diesel
exposure used in the analysis have differing degrees of specificity for DE. RSP is the
least specific for DE since it does not distinguish between organic and inorganic material.
AEM is expected to be the most specific for diesel exposure, since it excludes inorganic
material and adjusts for ETS. TEX, on the other hand, includes ETS and could possibly
correlate more closely with lung cancer risk than AEM since there is evidence that ETS
poses an independent lung cancer risk (EPA, 1990).
There is considerable uncertainty in making geographical distinctions in exposures.
The RRB provided an address for a representative of the last railroad for which a cohort
member worked, and this address was assumed to correspond to his work location
throughout his entire railroad employment. These addresses appeared to correspond
B-48 DRAFT-DO NOT QUOTE OR CITE
-------
reasonably well to the geographical locations suggested by many of the railroad names,
although no systematic study was made. A few of the larger railroads operate in more
diverse geographical areas and assigning a single location to these was problematic. For
these, the temperature adjustment may work reasonably well for railroads that operate
predominately East and West, but more poorly for those that operate predominately
North and South.
Given the importance of the exposure estimates to the quantification of risk and
the strong effect of temperature upon the exposures measured by Woskie et al. (1988), it
was considered important to take climatic differences into account in the exposure
estimates, insofar as possible. Despite their possible limitations, these estimates should
be at least as good, or better, than the estimates that were not climate-adjusted. As a
hedge against the possibility that climate-adjusted estimates were biased for some reason,
ARP, unadjusted for climate, was also used.
Dieselization of U.S. railroads was virtually complete (95%) by 1959, the first year
of follow-up of the cohort (Garshick et a/., 1988). However, a significant number of
diesel locomotives began to appear during the 1940s, with 14% of U.S. locomotives being
diescl by 1947 and 44% by 1951 (AAR, 1989). Moreover, practically the entire cohort
first worked for a railroad prior to 1950, and the majority prior to 1946. Thus, because
the cohort was only followed for 22 years, because the latency generally associated with
chemical or radiation-induced lung cancer is on the order of 20 years or greater, and
because early diesel locomotives were likely to have been more smoky than later models,
exposures prior to 1959 could be at least as important as exposures subsequent to 1959
in affecting lung cancer incidence in the cohort. Thus, it is important to quantify
exposures prior to 1959 even though there is limited information available from this time
period.
Exposure levels for a given year prior to 1959 were estimated by the product of
the estimated exposure level in 1959 and the percent of the U.S. railroad fleet that was
diesel in 1959. Thus, a person was assumed to have worked in the same job prior to
1959 as in 1959. More importantly, exposures were assumed to be equal among all
persons in a given job prior to 1959; in reality, those who worked with diesel equipment
probably experienced exposures as high (and probably higher, assuming early diesel
B-49 DRAFT-DO NOT QUOTE OR CITE
-------
locomotives were smokier and ventilation was poorer) as those occurring subsequent to
1959, while those who did not work with diesel equipment were unexposed. Although
the approach used in the analysis may result in average exposures during this period that
are approximately correct, by not having information on individual exposures, important
dose response information may have been obfuscated.
Potential for Confounding
Only sketchy information on smoking habits is available for this cohort. Garshick
et al. (1988) report that, in a survey of 517 current railroad workers, no difference was
found in the prevalence of smoking between workers with and without potential exposure
to DE. Also, a case control study of lung cancer, based on data provided by the RRB
(Garshick et al., 1987), found little difference between crude estimates of effect of diesel
exposure and estimates corrected for the effect of smoking.
ETS may also be confounded with DE as a potential cause of lung cancer. Albert
et al. (1983) compared the potency of DE from several types of engines to that of
cigarette smoke condensate in several short-term in vitro assays. DE was more potent
than cigarette smoke condensate in most, but not all, of these assays. However, there
were significant differences in potencies among DE from different types of engines.
Most of the data on diesel reported by Albert et al. came from experiments with small
diesel engines. Data from diesel railroad locomotives were not included in their analysis.
Table 3 suggests that concentrations of ETS smoke were comparable to
concentrations of DE in many jobs (the concentration of ETS is estimated as the
difference between RSP and ARP). If ETS and exhaust from diesel locomotives have
comparable potencies, then it would be difficult to separate their effect on lung cancer.
However, in that case, TEX, which is a marker for the combination of DE and ETS,
T
should provide a more consistent correlation with lung cancer than the other markers
considered
Evidence of an Effect of DE Upon Lung Cancer From This Study
Numerous analyses discussed elsewhere in this report on the relation between
various surrogate measures of diesel exposure and lung cancer mortality failed to
establish a relationship between exposure to DE and lung cancer. However, as discussed
above, the quantitative measures of diesel exposure used in this study have many
B-50 DRAFT-DO NOT QUOTE OR CITE
-------
potential shortcomings. Consequently, the negative finding may be due to weaknesses in
the exposure data rather than to an absence of diesel effect per se.
The principal evidence for an effect of DE upon lung cancer risk comes from
analyses in which diesel exposure was dichotomized (yes/no). One type of analysis
involved calculating the relative risk in exposed versus unexposed workers for six five-
year age intervals based on age in 1959. That analysis has now been conducted in four
different ways, two by Garshick et al. (1988) (their Tables 5 and 6) and two in the
current report (our Tables 5 and 6). These analyses give similar results. They show that
the relative risk increases progressively from the null value among workers aged 60-64 in
1959 to a statistically significant value of 1.4 among workers aged 40-44 in 1959.
In these analyses, workers assumed to be unexposed are those who worked as
clerks and signalmen in 1959. Both of these groups generally worked away from the
exhaust of operating trains. This designation of an unexposed group is generally
supported by the exposure estimates of Hammond et al (1988) (see also Table 3, which
was derived from the Hammond et al. data). Based on AEM, which is the marker that is
most specific for diesel exposure, clerks have the lowest estimated exposure (7.2 Mg/m3).
Although the exposure assigned to signalmen (23 Mg/m3) is higher than three of the
remaining ten groups, the estimated exposure of signalmen would appear to be lower
than the average estimated exposure over the three broad groupings (engineers and
firers; conductors, brakemen and hostlers; shop workers) used in the bulk of the analyses
reported herein.
Comparisons of age- and year-specific death rates for all deaths in this cohort with
corresponding rates for the general U.S. population strongly suggest that follow-up in this
cohort was incomplete beginning in about 1977 and became progressively more so with
each succeeding year. However, the higher rates among exposed workers do not appear
to be due to any lack of follow-up between 1977 and 1980 because restricting follow-up
through the end of 1976 does not change the pattern of relative risks appreciably
(Table 6).
On the one hand, this finding is reassuring in that it indicates that perhaps lack of
follow-up did not appreciably bias the findings in this study. On the other hand, the fact
that the increased risk persists even if follow-up ends in 1976 (the relative risk is higher
B-51 DRAFT-DO NOT QUOTE OR CITE
-------
in the group in the ages 40-45 in 1959 with follow-up ending in 1976 [1.64] than with
follow-up ending in 1980 [1.49]) seems at odds with the 20-year latency generally typical
of environmentally induced lung cancer.
When these analyses are performed separately for each category of exposed
workers (Table 7), it becomes evident that the observed pattern for the cohort is due
primarily to engineers and firers. The relative risks for this group follow the same
pattern as observed for the complete cohort, although the dependence on age in 1959 is
more pronounced for engineers and firers than for the complete cohort. By contrast,
although the relative risks among conductors, brakemen and hostlers are generally
elevated, they do not vary monotonically with age in 1959; rather, relative risks are higher
in the oldest and youngest groups and lowest among the intermediate age groups. The
relative risks of shop workers are not significantly elevated, either overall or in any age
group, and there is no obvious pattern of dependence of relative risk on age in 1959.
The pattern of relative risk observed in the different job groups has some features
that are not consistent with an effect of diesel exposure upon lung cancer rates in this
cohort. Exposures were estimated by Hammond et al. (1988) as being higher among
shop workers than among any other job group (see Table 3). This is consistent with the
fact that shop workers perform their duties near running locomotives in relatively
confined areas. Moreover, it is likely that early diesel shops were designed for steam
engines and did not ventilate DE efficiently. Woskie et al. (1988b) describe anecdotal
information and limited -historical data on nitrogen oxide and dioxide levels in diesel
shops that support higher levels in the shops in the 1950s and 1960s. Since the exposure
estimates obtained by Woskie et al. and Hammond et al. were not adjusted for
improvements in ventilation in diesel shops, the excess exposure experienced by the shop
workers over other groups was probably greater than estimated from the survey data.
There are several other features of the exposures of shop workers that need to be
considered. Some workers classified as shop workers by the ICC job codes worked in
non-diesel shops. However, Garshick et al. (1988) appear to have selected only job codes
involving substantial work around running locomotives. In this regard, they state, "...shop
workers had jobs mainly located inside the roundhouse and diesel repair facilities..."
B-52 DRAFT-DO NOT QUOTE OR CITE
-------
Shop workers were also the group most likely to be exposed to asbestos.
Although Garshick et al. (1988) did not find an association with potential for asbestos
exposure, any effect that was present would tend to increase the risk of shop workers,
which would make the lower risk in shop workers even more inconsistent with an effect
of diesel exposure in this group.
Thus, shop workers were both likely to have experienced the highest exposures to
DE and were the most likely to have been exposed to asbestos. The fact that no excess
lung cancer was observed among shop workers suggests that the excess observed among
groups exposed to lower levels of DE may not have been a result of exposure to DE.
Although an internal control group (clerks and signalmen in this study) should
generally be more appropriate than an external group such as U.S. white males, it is
possible that this may not be the case in an observational study such as this, particularly
since no information was available on smoking. Clerks, in particular, could differ from
engineers, firers, shop workers, etc., physically, socially, and in personal habits, including
use of tobacco. Thus, we believe that it was useful to complement the analysis based on
internal controls by considering external controls as well. It was this analysis that led us
to suspect that follow-up was incomplete in this cohort. In interpreting the comparisons
with U.S. white males in Table 5, it should be kept in mind that in addition to the
apparent lack of follow-up in the cohort, 11% of known deaths were not assigned a
specific cause.
The analysis of cumulative years of exposure was performed to compare to the
analysis of relative risk with years of exposure performed by Garshick et al. (1988). Both
analyses used the same intervals to characterize cumulative years of exposure. However,
whereas we used a Poisson regression model, the analysis by Garshick et al. (1988)
utilized a Cox regression model in which calendar year was used to define risk sets for
the partial likelihood and age was modelled as a continuous variable. The results of
these analyses were quite different; whereas Garshick et al. (1988) found relative risk to
be increasing with increasing years of exposure, we found a decreasing trend.
The reasons for these differences are not clear. Because of software limitations
we were not able to reproduce the Garshick et al. (1988) Cox regression analysis.
However, we were able to study elapsed time from 1959 until retirement or death in a
B-53 DRAFT-DO NOT QUOTE OR CITE
-------
Cox regression analysis. (Elapsed time coincides with cumulative years of exposure for
workers who did not have a break in their period of employment or change jobs.) This
analysis gave marginally significant increases in relative risk for each level of elapsed
time; however, no trend with elapsed time was apparent. The fact that we got the same
results for elapsed time in a Can regression analysis and a Poisson regression analysis
suggests that the differences between our results and those of Garshick et al. (1988) for
cumulative years of exposure are not due to differences between Cox regression and
Poisson regression, per se. Although the reason for these differences has not been
completely determined, it appears to be related to the fact that the three variables used
in the analysis - calendar year, age, and cumulative exposure - are likely to be all highly
correlated, which could make the results for cumulative exposure highly sensitive to the
specific methods used for controlling for age and time.
CONCLUSIONS
No relationship between measures of diesel exposure and lung cancer mortality is
demonstrated in this study. However, there are many limitations in using the data on
markers of diesel exposure collected by Woskie et al. (1988a) and Hammond et al. (1988)
to estimate exposures in the cohort of railroad workers studied by Garshick et al. (1988).
These limitations are potentially of sufficient magnitude to obscure any relationship
between exposure to DE and lung cancer that may exist in the cohort.
The higher relative risk of lung cancer among exposed workers relative to
unexposed workers reported by Garshick et al. (1988) was verified in the present study.
Related findings resulting from our work are as follows:
• The risk of lung cancer among engineers and firers was significantly
elevated relative to that of unexposed railroad workers, and the
variation in relative risk with age in 1959 (increasing risk with
decreasing age) is consistent with DE being responsible for the
observed excess.
B-54 DRAFT-DO NOT QUOTE OR CITE
-------
Although the risk of lung cancer among conductors, brakemen and
hostlers was significantly elevated relative to that of unexposed
railroad workers, the variation in risk with age in 1959 (higher at older
and younger ages than at intermediate ages) is not consistent with an
effect of DE
The risk of lung cancer among shop workers was not significantly
elevated relative to that of unexposed railroad workers. This is not
consistent with an effect of exposure to DE upon lung cancer risk
since it is likely that shop workers had significantly higher exposures to
DE than either of the other two groups of exposed workers.
Both internal evidence and comparisons with U.S. mortality rates
suggest that, beginning in about 1977, a significant number of deaths
occurring in this cohort went undetected.
In this study, follow-up of railroad workers mortality extends through 1980, which
is only 22 years from when dieselization of the U.S. railroads was essentially complete.
Since the time from first exposure until evidence of an increased risk of environmentally
induced lung cancer is often on the order of 20 years, the full impact of any effect of DE
upon lung cancer may not be captured by the current study. Moreover, there is evidence
that a sizable percent of the total deaths occurring in this cohort between 1977 and 1980
may not have been identified. Thus, it would be worthwhile to conduct a new study of
this cohort to take advantage of several additional years of follow-up now available. If
such a study » conducted, it is recommended that vital status be verified independently
of RRB records.
B-55 DRAFT-DO NOT QUOTE OR CITE
-------
Cox regression analysis. (Elapsed time coincides with cumulative years of exposure for
workers who did not have a break in their period of employment or change jobs.) This
analysis gave marginally significant increases in relative risk for each level of elapsed
time; however, no trend with elapsed time was apparent. The fact that we got the same
results for elapsed time in a Cox regression analysis and a Poisson regression analysis
suggests that the differences between our results and those of Garshick et al. (1988) for
cumulative years of exposure are not due to differences between Cox regression and
Poisson regression, per se. Although the reason for these differences has not been
completely determined, it appears to be related to the fact that the three variables used
in the analysis - calendar year, age, and cumulative exposure - are likely to be all highly
correlated, which could make the results for cumulative exposure highly sensitive to the
specific methods used for controlling for age and time.
CONCLUSIONS
No relationship between measures of diesel exposure and lung cancer mortality is
demonstrated in this study. However, there are many limitations in using the data on
markers of diesel exposure collected by Woskie et al. (1988a) and Hammond et al. (1988)
to estimate exposures in the cohort of railroad workers studied by Garshick et al. (1988).
These limitations are potentially of sufficient magnitude to obscure any relationship
between exposure to DE and lung cancer that may exist in the cohort.
The higher relative risk of lung cancer among exposed workers relative to
unexposed workers reported by Garshick et al. (1988) was verified in the present study.
Related findings resulting from our work are as follows:
• The risk of lung cancer among engineers and firers was significantly
elevated relative to that of unexposed railroad workers, and the
variation in relative risk with age in 1959 (increasing risk with
decreasing age) is consistent with DE being responsible for the
observed excess.
B-54 DRAFT-DO NOT QUOTE OR CITE
-------
Although the risk of lung cancer among conductors, brakemen and
hostlers was significantly elevated relative to that of unexposed
railroad workers, the variation in risk with age in 1959 (higher at older
and younger ages than at intermediate ages) is not consistent with an
effect of DE
The risk of lung cancer among shop workers was not significantly
elevated relative to that of unexposed railroad workers. This is not
consistent with an effect of exposure to DE upon lung cancer risk
since it is likely that shop workers had significantly higher exposures to
DE than either of the other two groups of exposed workers.
Both internal evidence and comparisons with U.S. mortality rates
suggest that, beginning in about 1977, a significant number of deaths
occurring in this cohort went undetected.
In this study, follow-up of railroad workers mortality extends through 1980, which
is only 22 years from when dieselization of the U.S. railroads was essentially complete.
Since the time from first exposure until evidence of an increased risk of environmentally
induced lung cancer is often on the order of 20 years, the full impact of any effect of DE
upon lung cancer may not be captured by the current study. Moreover, there is evidence
that a sizable percent of the total deaths occurring in this cohort between 1977 and 1980
may not have been identified. Thus, it would be worthwhile to conduct a new study of
this cohort to take advantage of several additional years of follow-up now available. If
such a study is conducted, it is recommended that vital status be verified independently
of RRB records.
B-55 DRAFT-DO NOT QUOTE OR CITE
-------
REFERENCES
Albert R, Chen C 1986. U.S. EPA diesel studies on inhalation hazards. In: Carcinogenkity and
Muugenicity of Diesel Engine Exhaust Ishinishi N, et aL, eds. Elsevier, New York. pp. 411-419.
Albert RE, Lewtas J, Nesnow S, Thorslund TW, Anderson E 1983. Comparative potency method for
cancer risk assessment: Application to diesel paniculate emissions. Risk Anal 3:101-117.
Association of American Railroads (AAR). 1989. Railroad Facts. AAR, Washington, DC
BEIR IV. 1988. Health Risks of Radon and Other Internally Deposited Alpha-emitters. Report of the
Committee on the Biological Effects of Ionizing Radiation. National Academy of Sciences, Washington,
BEIR V. 1990. Health Effects of Exposure to Low Levels of Ionizing Radiation. Report of the
Committee on the Biological Effects of Ionizing Radiation. National Academy of Sciences, Washington,
DC
Cox DR. 1972. Regression models and life tables (with discussion). J R Statist Soc [Ser B] 34:187.220.
Cox DR, Undley DV. 1974. Theoretical Statistics. Chapman and Hall, London.
Crump K. 1984. An improved procedure for low-dose carcinogenic risk assessment from animal data. J
Environ Pathol Toxicol 5:339-348.
Crump K, Allen B. 1985. Methods tor quantitative risk assessment using occupational studies. Amer Stat
39(Pt2):442-450.
Crump K, Howe R. 1984. The multistage model with a time-dependent dose pattern: applications to
carcinogenic risk assessment Risk Anal 4:163-176.
Cuddihy RG, Griffith WC, Mcdellan RO. 1984. Health risks from light duty diesel vehicles. Environ Sci
Technol 1&14A-21A.
Environmental Protection Agency (EPA). 1990. Health effects of passive smoking: Assessment of lung
cancer in adults and respiratory disorders in children. EPA/600/6-90/D06A. Office of Health and
Environmental Assessment, Office of Atmospheric and Indoor Air Programs, Washington, DC
Ferguson BV. 1990. Personal communication. Director, Bureau of Research and Employment Accounts,
Railroad Retirement Board, Chicago, IL.
Garshick E 1991. Personal communication. Brockton Veterans Administration Medical Center,
Brockton, MA.
Garshick E, Scbenker M, Junoz A, et aL 1987. A case-control study of lung cancer and diesel exhaust
exposure in railroad workers. Am Rev Respir Dis 135:1242-1248.
Garshick E Scbenker M, Munoz A, et al. 1988. A retrospective cohort study of lung cancer and diesel
exhaust exposure in railroad workers. Am Rev Respir Dis 137:820-825.
Hall NEL, Wynder EL 1984. Diesel exhaust exposure and lung cancer. A case-control study. Environ
Res 34:77-86.
B-56 DRAFT-DO NOT QUOTE OR CITE
-------
Hammond S, Smith T, Woskie S, Leaderer B, Bellinger N. 1988. Markers of exposure 10 diesel exhaust
and cigarette smoke in railroad workers. Am lod Hyg Assoc J 4W16-522.
Harris J. 1983 Diesel emissions and lung cancer. Risk Anal 3:83-100.
Howe G, Fraser D, Lindsay J, Presnal B, Yu S. 1983. Cancer mortality (1965-1977) in relation to diesel
fume and coal exposure in a cohort of retired railway workers. J Nail Cancer Inst 70:1015-1020.
Interstate Commerce Commission (ICC). 1951. Bureau of Transport Economics and Statistics. List of
occupations or positions in railroad service, file 21-B-2. USGPO, Washington, DC
Kaplan L 1959. Relationships of noxious gases to carcinoma of the lung in railroad workers. J Am Med
Assoc 171:2039-2043.
Lewtas J, Williams K. 1986. A retrospective view of the value of short-term genetic bioassays in
predicting the chronic effects of diesel soot. In: Carcinogenicity and Mutagenicity of Diesel Engine
Exhaust Ishinishi N, et aL, eds. Elsevier, New York. pp. 119-140.
Mauderly JL, Jones RK, McCteUan RO, et al. 1986. Carcinogenicity of diesel exhaust inhaled chronically
by rats. In: Carcinogenicity and Mutagenicity of Diesel Engine Exhaust Ishinishi N, et al., eds. Elsevier,
New York. pp. 397-410.
McClellan RO. 1987. Health effects of exposure to diesel exhaust particles. Ann Rev Phannacol Toxkol
27:279-300.
Morrow PE 1988. Possible mechanisms to explain dust overloading of the lungs. Fund Appl Toxicol
10-.369-384.
National Oceanic and Atmospheric Administration (NOAA). 1978. Climate of the States. No. 60.
NOAA, Asheville, NC
Oliver JE. 1987. The Encyclopedia of Climatology. Van Nostrand Reinhold Company, New York. p.
833.
Preston DL, Lubin JH, Pierce DA. 1990. EPICURE. Programs for Regression Analyses of
Epidemiologic Data Including: AMFTT, PECAN, G'MBO, PEANUTS and DATAB.
Smith R, Stayner L 1990. An exploratory assessment of the risk of lung cancer associated with exposure
to diesel exhaust based on a study in rats. Division of Standards Development and Technology Transfer,
National Institute for Occupational Safety and Health.
Waller R. 1980. Trends in lung cancer in London in relation to exposure to diesel fumes. In: Health
Effects of Diesel Engine Emissions. Vols 1 and 2. Pepelko W, et aL, eds. U.S. Environmental Protection
Agency, Cindnaati, OH. pp. 1085-1097.
Waxweiler R, Wagoner J, Archer V. 1973. Mortality of potash workers. J Occup Med 15:486-489.
Wong O, Morgan R, Keifets L, et aL 1985. Mortality among members of a construction equipment
operators union with potential exposure to diesel exhaust emissions. Br J Ind Med 42:435-448.
Woskie SR, Smith TJ, Hammond SK, Schenker MB, Garshick E, Speizer FE. 1988a. Estimation of the
diesel exhaust exposures of railroad workers: L Current exposures. Am J Ind Med 13:381-394.
B-57 DRAFT-DO NOT QUOTE OR CITE
-------
Woskie SR, Smith TJ, Hammond SK, Scheoker MB, Garshick E, Speizer FE 1988b. Estimation of the
diesel exhaust exposures of railroad workers: IL National and historical exposures. Am J tod Med
13:395-404.
B-58 DRAFT-DO NOT QUOTE OR CFTE
-------
APPENDIX
GRAPHS OF RELATIVE RISK FOR VARIOUS EXPOSURE
MEASURES AND SUBSETS OF THE COHORT
-------
This the appendix contains figures for the risk assessment presented in Appendix
B. In these figures, relative risk is plotted against the average dose for several dose
categories. Figures are provided for each of the five markers for DE and for exposure to
DE accumulated over four time periods: more than three years in the past, from four to
eight years in the past, from nine to 18 years in the past, and more than 18 years in the
past. Figures are provided for the complete cohort and for engineers/firers,
conductors/brakemen, and shop workers separately.
APP-1 DRAFT-DO NOT QUOTE OR CITE
-------
c. -e 5
ii . . \
T3 • > " T .. ...^ i
O * 5
£ li "?1
«*i» re • *
3 5i -?
Q) ! -»
Li. r
I-
fi-
APP-2 DRAFT-DO NOT QUOTE OR CITE
-------
0
LL
-i
0
S!
£ 5
f I .
=
1
fi
i
APP-3 DRAFT-DO NOT QUOTE OR CITE
-------
-§
CO ]
1
s\ §i
9 9
P ll I
C I e*
= I "S!
« ,
I i
1
>
APP-4 DRAFT-DO NOT QUOTE OR CITE
-------
* I I
:? i-1 s!
< *•• '—' i
P i! i
re s s
°\
L
m «
e o
APP-5 DRAFT-DO NOT QUOTE OR CITE
-------
c. «
< 1! —. i
-------
CD
0
D
Q)
LL
t
E
<
APP-7 DRAFT-DO NOT QUOTE OR CITE
-------
=
D)
LL j I
i
o a a e o e
•to •*,•»*
APP-8 DRAFT-DO NOT QUOTE OR CITE
-------
! i
J
""" E >
<<: • ' :
SJ .!
°
0
D)
LL
APP-9 DRAFT-DO NOT QUOTE OR CITE
-------
si i
-------
-I
-I
c « t
< '
31 'I
*; fs
i . . &
51 }
O) I
II
I - 8
> IN
-• s
i
i
APP-11 DRAFT-DO NOT QUOTE OR CITE
-------
LL
•o : i
£ J . t
£ fj ""I
—i 5 j I
i !
APP-12 DRAFT-DO NOT QUOTE OR CITE
-------
-I
C\J
0
D
L
!!
fi
=> i
•
i
u
- S
- s
APP-13 DRAFT-DO NOT QUOTE OR CITE
-------
s
CO I
ll
9> II . I
§> "I ~ -.'
LL I
i
APP-14 DRAFT-DO NOT QUOTE OR CITE
-------
0)
D
D)
U_
I
a
1!
-• t
1
j
i
APP-15 DRAFT-DO NOT QUOTE OR CITE
-------
10
• - • -
< i! I
fi >
!
= ' - ' "I
O) I J
LL i . _ \-
APP-16 DRAFT-DO NOT QUOTE OR CITE
-------
C\J
0)
L
ii
I!
2 J
f
-8
]
-I
APP-17 DRAFT-DO NOT QUOTE OR CITE
-------
-8
C\J
c\i
o
,1
< *
^ £ m . " • '"•••
«> i
m * «— .
ll
g>
u- i |.,
- 8
if - 8
e
APP-18 DRAFT-DO NOT QUOTE OR CITE
-------
-8
00 A •—• -i
• & I
C\J
- 5*
0 !i !
'\
O)
LL ' *
i-*
- 8
e e e e a e
APP-19 DRAFT-DO NOT QUOTE OR CITE
-------
• — — • -I
•I
£
j
• Q. » ~ §
C\J Si I
<-0 f
*:•
LL S
1 i
I
i
APP-20 DRAFT-DO NOT QUOTE OR CITE
-------
IT)
CO
0
O>
LL
i
II
t!
re ;
b
-!
si
|
s-
APP-21 DRAFT-DO NOT QUOTE OR CITE
-------
CO
Q. i
CM 5!
< \\ i
0 !! ' ' ~sl
LL ! • •
i
0) ° .1
~ e
i
APP-22 DRAFT-DO NOT QUOTE OR CITE
-------
LL
d V
< J
l »
ft
-------
00 l
™- i!
Is
a? f!
i_ i i
rr\ ° *
~ S
LL J
) ,
!
e o e o e e e
APP-24 DRAFT-DO NOT QUOTE OR CITE
-------
O> i
cvi |l
o f!
o> ";
u_
.
-
*
. .1
s
i
° ?
^
£
.1
°i
1
.1
e
CM
e
e
APP-25 DRAFT-DO NOT QUOTE OR CITE
-------
D)
LL
5
• *
I I
j
£ «l
—^ I i - :l
-•
i
I
" |
i
APP-26 DRAFT-DO NOT QUOTE OR CITE
-------
LL I !
M )
CVJ 5
2' s
*i :f
Si !
is -i
D) °i ..*
APP-27 DRAFT-DO NOT QUOTE OR CITE
-------
C\J I
T- ai
rvi <«
C\J si ••'
|| |
0 ll *
O)
LL
w
5
I
APP-28 DRAFT-DO NOT QUOTE OR CITE
-------
CO
c\i
0)
g>
L.
s
i
I •:
S-5
N e
a
APP-29 DRAFT-DO NOT QUOTE OR CITE
-------
1
t\l
0)
O)
LL
!!
o |
i i-
l . . f
3 • 'I
5-
i
r» - • • » « O
- o o e o
»»a ».JM«u
i
J
e
e
e
APP-30 DRAFT-DO NOT QUOTE OR CITE
-------
C\J
O I I
g> J . - . i
LL i 1-:
• - . I
J
u
r-
i
APP-31 DRAFT-DO NOT QUOTE OR CITE
-------
CO
?*
i
I
O) °
r1
• < •
- i
I
I - R
I
APP-32 DRAFT-DO NOT QUOTE OR CITE
-------
CM
CO
Q)
D)
LL
.1
!J
f 3
« i
li
u
-1
t
E
i
§]
1
- 8
I
I '
APP-33 DRAFT-DO NOT QUOTE OR CITE
-------
CO
CO
0
O)
LL
•
- 8
- 6
i
"i
!
- 9
!i
eoeeeeo
APP-34 DRAFT-DO NOT QUOTE OR CITE
-------
-I
-•
CO
0
O)
LL
}
f j
l|
o J
•
- i
i
.«!
"i
. i
- s
1
• k
- 9
APP-35 DRAFT-DO NOT QUOTE OR CITE
-------
-I
in
CO
<
0)
o
LL
2
UJ
i
i
- 8
- S
APP-36 DRAFT-DO NOT QUOTE OR CITE
-------
CO
<
CD
f
3 I! I
° *
-8
e e
APP-37 DRAFT-DO NOT QUOTE OR CITE
-------
CO
0
L
2
UJ
< l
-8
s-
e e e
APP-38 DRAFT-DO NOT QUOTE OR CITE
-------
rl
00
CO
0)
g>
LL
1
i
-2
i
I
!
e o
APP-39 DRAFT-DO NOT QUOTE OR CITE
-------
05 i f
CO . i
*
• S
£ 5 I
3 H '
g> °|
LL " _ j:'
v
$
- i
i
APP-40 DRAFT-DO NOT QUOTE OR CITE
-------
o
^^^
CO
^^^_
0
M_
O)
• MBB
LL
*
[
B
2l
< i
11
i!
.E i
Di
w
J
|
u
<
1
I
i
1 .
I •
i
* M «•
-!
i
f
a
p *
I
«
- s
->
,
» • * ft O
o o e e
APP-41 DRAFT-DO NOT QUOTE OR CITE
-------
-8
• i •
CO
<
D
LL
^U B •
$] «
« • ^
M j
0 |
I 1
(j S
1 - «
1 .
y •— _ ••«
i
• •'•-••vine
- o o a e ;
APP-42 DRAFT-EX3 NOT QUOTE OR CITE
-------
C\J
CO
0
D
D)
LL
s
2 I
3 I
5i
E
a
c
- 8
•
- SI
APP-43 DRAFT-DO NOT QUOTE OR CITE
-------
co i
-8
UJ J
CO |i
IS ,1
V n • 8 *
fl} » - • • ^
s II !
O)
E I ' ~ I
1-1
s
APP-44 DRAFT-DO NOT QUOTE OR CITE
-------
-------
-§
*^
CO
0
D
D)
• MM^B
LJL
^ |
Si
11
fl
II
"i
5
1
-S
I
~ 8i
!
-8
—
i:
• 18
' Is
APP-46 DRAFT-DO NOT QUOTE OR CITE
-------
0
0.1
g;
i <
a* •>
L
-si
£
«
C
i
-i
1
APP-47 DRAFT-DO NOT QUOTE OR CITE
-------
C\J
0
O>
LL
J
i
I!
o
1
APP-48 DRAFT-DO NOT QUOTE OR CITE
-------
CO
•
«fr
<
CD
•5
O>
LL
ii
I
i
•
e e e e e e
APP-49 DRAFT-DO NOT QUOTE OR CITE
-------
0)
D)
LL
0.1
c/3 a
o J
1;
APP-50 DRAFT-DO NOT QUOTE OR CITE
-------
If)
0)
L
a
c/5 5
C a
fl
n >*
i!
;=!
..I
APP-51 DRAFT-DO NOT QUOTE OR CITE
-------
CD
i* •
li *
o ?i — "i
*- Si - *
^^^M PV * •
g> '! -I
u. I - | '
1-5
w
I;1
i
e
e
APP-52 DRAFT-DO NOT QUOTE OR CITE
-------
0)
D
O>
LL
I
1!
o o e
APP-53 DRAFT-DO NOT QUOTE OR CITE
-------
op
0)
O)
LL
ii
ii
APP-54 DRAFT-DO NOT QUOTE OR CITE
-------
0> j
0)
< li
i
I
D *
;Z *
g) °j
LL u
•
V
I —
J
• • • « o
c e e e
APP-55 DRAFT-DO NOT QUOTE OR CITE
-------
LL
1
**•
I! -I
«-. I
ii ' '• I
, <
e
APP-56 DRAFT-DO NOT QUOTE OR CITE
-------
0
L
Climate Adjusted RSP
CunMMM* E«po«H« «•>!*»• llu* !• VMII ki •<• P«l In I
MMitf M% co^M*«e* toMl fcH^nrtli
i 1
i : '
&
-!
i
-1
«
~ e
M
"" e
*
-•••NO
e a e e
APP-57 DRAFT-DO NOT QUOTE OR CITE
-------
:.,
•r—
•
^j-
<
0)
3
O>
LL
a
Q. 1
(/) *
cc i
S s
il
JP •
< i
fi
3i
1
•4
i
s
i
i
i
'!
e e e o
APP-58 DRAFT-DO NOT QUOTE OR CITE
-------
CO
0)
D)
LL
1
ftf
= i
Is
< j
f i
3i
Uf
'
•
1
1
1
:
-
• * .
-
. . 1
• ' • J
i-
i
i
n • M • <• a
a o
j
«
E
{
1
I
5
*
Q
APP-59 DRAFT-DO NOT QUOTE OR CITE
-------
0 a
\^
=
O)
-•. c
*
''
,-
i'
APP-60 DRAFT-DO NOT QUOTE OR CITE
-------
CD
L
i
1
ll
I!
fi
I
i
i
I
.-
r
APP-61 DRAFT-DO NOT QUOTE OR CITE
-------
-8
in si i
j
CD
-------
-I
C\J
LO
0>
L
x £
UJ i
Climale Adjusted '
CwnUMM* E >po*uw m •• l>nl 4 lo • VMI
,
»
i
1
•
W
s
ii
M - • • «
- e o e
I *
i
i
*!
e
APP-63 DRAFT-DO NOT QUOTE OR CITE
-------
-8
CO
•
L^}
^
0
D
D)
LL
i
x 1
i 1 1 i ,r
••* v
»- a
i;
S 5
fi
•I
tW
1
•
i
i ;
i i
1 i
- s
- 8
;
- §
1 '
1:
l-§
§i
SLS
l!
ii
APP-64 DRAFT-DO NOT QUOTE OR CITE
-------
LO
0)
ID
L
X ,
-
1
!
2
c
-s
m
i
1-1
X
APP-65 DRAFT-DO NOT QUOTE OR CITE
-------
-8
IT)
LT>
CD
LL
i
o
-8
8}
I
]
s
APP-66 DRAFT-DO NOT QUOTE OR CITE
-------
CD
IT)
CD
ID
O>
LL
1
UJ S
*!
1
S I
P8 6
-8
6
«•
- 8
APP-67 DRAFT-DO NOT QUOTE OR CITE
-------
•I
• X I
If) |!
^^ § e
'•Sf i
CD fi
^•B |5 f
§> si
• •••§ •
Li- 1
O
j
1
1
1
I
|
i
i
?
i
•
i
3
.1
- 8 c
i
}
J
*• 8
-
i
^
^
e o e e e e e
APP-68 DRAFT-DO NOT QUOTE OR CITE
-------
CO
LO
<
0
D
D>
U_
s
xf
g
t in
le
m tii
-§
»|m»im
-2
tr*
APP-69 DRAFT-DO NOT QUOTE OR CITE
-------
- s
o> !
• x :
IT) Si
J
I
s
i
e e e e
APP-70 DRAFT-DO NOT QUOTE OR CITE
-------
in
0
D
O>
LL
x 1
LU e
Climate Ac
ipMui* tat Mw« Ilwn
•
J
-I
i
-
APP-71 DRAFT-DO NOT QUOTE OR CITE
-------
IT)
0)
L
si
11
u s
-s
-I
APP-72 DRAFT-DO NOT QUOTE OR CITE
-------
C\J
X
PMI to Stop
E
LO
<
o
D
O)
LL
Climate Adjusted I
:MHUMM E ipMun ta> MM* than If ¥*••«•
4
i's
j
o e
APP-73 DRAFT-DO NOT QUOTE OR CITE
-------
CO
in
0
D
r^\
&
*l
LU 1
1 «
- !
1 *
i 3
0 $
~ o —
• • e
!1
1
• a
f
e
1
LL
1
s
|
• • v M e
e e e e
APP-74 DRAFT-DO NOT QUOTE OR CITE
-------
ID
0
D)
LL
2 {
|i
si
Ii
rZi
I
J I
II.
sr
II
APP-75 DRAFT-DO NOT QUOTE OR CITE
-------
in
in
CD
Z5
D)
LL
i
s
>• i
i
APP-76 DRAFT-DO NOT QUOTE OR CITE
-------
ESTIMATES OF INCREASE IN RELATIVE RISKS IN LUNG CANCER
BASED UPON PUBLISHED RISK ASSESSMENTS FOR DIESEL EXHAUST
Harris (1983) conducted a risk assessment for diesel paniculate using data from a
study of London Transport Workers (Waller, 1980). Harris assumed a relative risk of
lung cancer resulting from exposure to diesel paniculate of the form (1 + 8 * X), where
X is the excess cumulative exposure to diesel emissions in Mg/m3-years, and e is the
potency of diesel paniculate measured as the increase in relative risk per Mg/m'-years
exposure. Harris estimated e = 1.23x10"* per Mg/m3-year (s.e. * 1.86xlO"*). This
estimate was derived from a negative study, as evidenced by the fact that the standard
error exceeds the estimate.
Harris also estimated e using a relative potency approach that utilized
epidemiological data on coke oven and roofing tar workers, and data from short-term
bioassays of extracts of coke oven emissions, roofing tar emissions, and diesel emissions
from three automobile engines and a Caterpillar engine. The overall mean estimate of
the potency obtained from extracts from the three automobile engines was e = 0.35x10"*
per /jg/m'-year. The estimates obtained from the Caterpillar engine ranged from
e = 0.01x10"* to 6 « 0.25x10"* per jig/m'-year.
Smith and Stayner (1990) applied a multistage time-to-tumor model (Crump and
Howe, 1984) to the data from a study of rats exposed throughout life to DE (Mauderh/
et al., 1986). They obtained estimates, for the unit risk (additional lifetime risk from
exposure to 1 Mg/m3) in the range of IxlO'5 to 2xlO~5. Their estimate assumes
occupational exposure from age 18 through age 65 (47 years).
To convert this estimate of lifetime risk made from the Mauderh/ et al. animal
study into estimates of e, a linear dose model for the relative risk of the form
RR = 1 + e * D,.j,
was assumed, where D,.5 is cumulative exposure through five years prior to the age, t, of
observation. The lifetime risk of death from lung cancer in a person occupationally
exposed to 1 jig/mj from age 18 through age 65 is estimated based on mortality rates in
APP-77 DRAFT-DO NOT QUOTE OR CITE
-------
1971 for U.S. males by five-year age intervals for total deaths and lung cancer (1971
corresponds to the midpoint of the years of follow-up of the Garshick et al (1988)
cohort). The rates for lung cancer were modified by multiplication by the relative risk
(RR) calculated using the appropriate exposure, D. The rates for all deaths are similarly
modified to account for the increase in lung cancer death rates. A life table approach
(Crump and Allen, 1985) was then applied to the modified rates to calculate the lifetime
probability of death from lung cancer. The additional risk of lung cancer from exposure
to diesel paniculate was calculated by subtracting from this the corresponding risk of
death from lung cancer based on the unmodified death rates.
This procedure was applied for different values of 6 until a range of values for &
were found that corresponded to the range of lifetime risks obtained from the Mauderty
el al. study by Smith and Stayner. This range was 4. 6x10"* to 9.2x10"*, as listed in
Table 10.
Albert and Chen (1986) applied the quantal multistage model (Crump, 1984) to
pre-publication data from the same study by Mauderty et al and obtained a 95% upper
limit of 1.2xlO"$ per Mg/m3, corresponding to the exposure pattern in the animal study
(seven hours per day, five days per week for life). To convert this estimate so that it
corresponds to 47 years of exposure eight hours per day in humans, the following
adjustment was applied:
(1.2xlO"J) • (8 hours/7 hours) * (47 years/75 years) * 0.86xlO's per
(see also Smith and Stayner, 1990). The corresponding estimate based on the MLE of
potency from the animal data, rather than the 95% upper limit, is 0.26x10 5 per
/ig/m'-year. These estimates of lifetime risk were converted into estimates of 6 in the
same way as the estimates made by Smith and Stayner (1990). The results are displayed
in Table 10.
Albert and Chen also obtained an estimate of lifetime risk based on a comparative
potency method that was in the same units as the estimate obtained from the Mauderty
APP-78 DRAFT-DO NOT QUOTE OR CITE
-------
et al data. Applying the same procedure to this estimate yielded an estimate of
6 = S.OxlO"6 per Mg/m3-year.
After reviewing epidemiological data on lung cancer from exposure to several
sources of poh/cyclic aromatic hydrocarbons (PAHs) and short-term assay data on these
extracts from these same exposure sources, Cuddihy et al. (1984) concluded that the "lung
cancer risk from exposure to DE is not likely to result in more than 0.1 lung cancer
deaths each year in a population of 100,000 people per Mg/mV It was assumed that this
risk corresponds to 7.5xl05 lifetime risk (yearly risk times 75-year life span). Using the
same life table approach that was appb'ed in Smith and Stayner's estimate, this estimate
corresponds to a relative risk of 8 * 0.46x10"* per /jg/m'-year.
APP-79 DRAFT-DO NOT QUOTE OR CITE
-------
APPENDIX C
ALTERNATIVE MODEL FOR DIESEL CANCER RISK
ASSESSMENT
December 1994 DRAFT-DO NOT QUOTE OR CITE
-------
i APPENDIX C
2 ALTERNATIVE MODEL FOR DIESEL CANCER
3 RISK ASSESSMENT
4
5 Cl. INTRODUCTION
6 As previously discussed in Chapter 11, the most appropriate method to assess
7 cancer risk of diesel exhaust is to take into account effects of both particles (carbon core)
8 and organics because evidence exists that both agents are involved in carcinogenic
9 process. The reasons for this conclusion are based on the following observations: (1)
10 organics include a variety of polycyclic aromatic hydrocarbons (PAHs) and
11 nitroaromatics, many of which are known to be carcinogenic; (2) the results of recent
12 studies on inert particles and carbon black in rats strongly support the hypothesis that the
13 carbon core of the diesel panicle may be the primary component responsible for the
14 induction of lung cancer; (3) PAHs are unlikely responsible for all observed tumors
15 because they account for less than 0.1 Mg/mg paniculate matter (Tong and Karasek,
16 1984); and (4) the observation of disproportionate high tumor incidence in high exposure
17 animals coincides with disproportionate increase of cumulative lung burden of diesel
18 panicle as exposure concentration increases.
19 A workshop on Research Needs for Risk Assessment of Inhaled Paniculate Matter
20 was organized and sponsored by the U.S. Environmental Protection Agency (EPA) in
21 March, 1992. Tne purpose of the workshop was to determine the extent of information
22 that can be used for quantitative risk assessment and to discuss mechanisms of
23 panicle-induced lung tumors to serve as a guidance for future research needs. Two
24 major, among several other, conclusions that are relevant to quantitative risk assessment
25 were reached by the Workshop:
26
27 (1) panicle overloading of the lung tissue may induce both initiation (by PAH
28 specific adducts and adducts through oxygen radicals) and cell proliferation
29 steps in tumor formation, and
30
31 (2) more research is needed to improve the risk assessment of panicle-induced
32 lung cancers.
C-l DRAFT-DO NOT QUOTE OR CITE
-------
1 Although there are not enough data available to construct a biologically based
2 dose-response model, it is desirable to investigate implications of the hypothetical
3 mechanisms proposed by the workshop. The purpose of the alternative modeling
4 presented in this report is to do just that. Briefly, the biological issues and their
5 implications to quantitative risk assessment that we would like to consider are the
6 following.
7
8 1. Particles deposited in lung are phagocytized by alveolar macrophages. Because the
9 phagocytizing macrophages in animals from high-dose group may be more likely to
10 be activated to release mediators including reactive oxygen species, cytokines, and
11 growth factors, it is of interest to determine whether or not the available tumor
12 response data are consistent with the hypothesis that the particle burden affects
13 both initiation and proliferation in carcinogenic process.
14
15 2. Organic materials can also induce specific adducts which may contribute to cell
16 initiation. However, given its low content, the contribution of organics to tumor
17 induction may be very small. Can a dose-response model that is consistent to the
18 proposed biological concept be constructed with both organics and particles as dose
19 metrics?
20
21 3. If a model that has the above biological interpretation and is consistent with the
22 bioassay data can be constructed, what would be its implications on quantitative risk
23 assessment of diesel exhaust emissions, and how would its results compare with
24 those predicted by the linearized multistage (IMS) model?
25
26
27 C2. PRELIMINARY CONSIDERATIONS
28 In order to evaluate the impact of various biological assumptions on diesel risk
29 assessment, it is necessary to construct a mathematical dose-response model that takes
30 into account the biological mechanisms discussed in the EPA workshop. Because an
31 issue of significant importance in diesel risk assessment is the effect of lung overloading
32 on tumor induction, the model should possess the fol/lowing properties.
33
34 1. It should depend on both types of dose metrics: organics, and carbon core.
35 It should allow one to account for the contribution of organics and carbon core
36 individually and/or jointly to tumor induction/formation.
37
C-2 DRAFT-DO NOT QUOTE OR CITE
-------
1 2. It should allow for the possibility that model parameters can change with time
2 because of the increasing lung burden during exposure.
3
4 3. The cell proliferation and tumor induction/formation should be stochastic in nature;
5 it is not realistic to assume deterministic clonal growth. For instance, it should not
6 be required to assume that all cells divide at the same age.
7
8 To accomplish these goals, we assume that a normal cell can be initiated by both
9 organics and carbon core. Denote the initiation rate by ^j, which is a function of
10 background and diesel-induced rates (as specified below). Because an initiated cell (I-
11 cell) eventually either goes into cell death, or enters the cell cycle (including cells in
12 quiescence, GQ), it is reasonable to assume that the cell lifetime for an I-cell follows
13 certain probability distribution. Under this model, a cell in G0 phase is equivalent to the
14 case where it has a very long lifetime with certain probability (i.e., in the right-hand tail
15 of the cell lifetime distribution). At the end of an I-celTs lifetime, it either dies (death)
16 with probability ft, divides into two daughter cells (birth) with probability a, or divides
17 into one I-cell and one malignant cell (second transition) with probability ^ a + ft + V>i
18 = 1. Instead of assuming that a single malignant cell is equivalent to a tumor as in the
19 MVK model proposed by Moolgavarkar and colleagues (1979, 1981), we assume that a
20 malignant eel] has a certain probability to become a tumor; this probability is assumed
21 dose-dependent, thus allowing for an evaluation of dose effect on tumor progression. It
22 should be noted that the proposed model does not exclude the possibility that it may
«*
23 take more than one step (for a normal cell) to become "initiated". The rate of initiation
24 used in the model should be viewed as a net rate which represents several genetic
25 alterations and repairs.
26
27
28 C3. MATHEMATICAL MODEL AND PARAMETERS ESTIMATION
29 We shall proceed to construct a dose-response function P(t:d.D), probability of
30 cancer by time (age) t, which depends on both organic, d, and particle (carbon core), D,
31 and incorporates the biological concept discussed previously. Because the model
32 parameters that are not directly observed in laboratory can only be statistically estimated
33 from high concentration cancer bioassay data, the model constructed should not be
C-3 DRAFT-DO NOT QUOTE OR CITE
-------
1 considered a valid model of diesel-induced carcinogenesis; uncertainty about the low-dose
2 extrapolation still remains. Some discussions about the need for further laboratory
3 measurements will be given later.
4 The model with the desirable features discussed above falls into one of several
5 classes of stochastic models that have been developed by EPA's Office of Health and
6 Environmental Assessment (OHEA): namely, the stochastic model which was originally
7 proposed by Chen and Farland (1991) and extended into one with time varying
8 parameters by Tan and Chen (1992). This model will be used as the basis for
9 constructing a biologically based dose-response model. A brief mathematical description
10 is presented in Appendix C-2.
11 The time to event data from Mauderly et al. (1987) are used to estimate model
12 parameters. The data from Mauderly et al. are useful because they contain information
13 on natural mortality and serial sacrifice of animals with or without (malignant) tumors,
14 valuable information for estimating tumor latency. To utilize the information from serial
15 sacrifice in Mauderly et al. an (E-M) algorithm is derived (see Appendix C-l) and used
16 to calculate maximum likelihood estimates of parameters.
17
18
19 C3.1 Model Parameters and Notations
20 The following parameters are incorporated in the dose-response model, which
21 includes initiation rate (MJ), proliferation rate (ya), conversion rate (y^X and
22 probability of progression (q). The death rate for the initiated cells is implicitly defined
23 by y(l - ji2 ~ °)- These parameters are all dose dependent.
24
2
25 D: Dose of carbon core, mg/cm of lung epithelial surface; D varies over time
26
27 d: Dose of organics, mg/cm of lung epithelial surface
28
29 /*j: Dose-related initiation rate (per cell per day) that depends on MQ (background
30 rate), d, and D by MI = MoO + acl + *>D); a and b are paramaters to be estimated.
31
32 M2: Probability of producing a malignant cell at the end of an initiated cell (I-cell)
33 lifetime
34
C-4 DRAFT-DO NOT QUOTE OR CITE
-------
1 a: The probability that an I-cell divides into two daughter cells at the end of its
2 lifetime
3
4 q: Probability that a single malignant cell will develop into a malignant tumor
5
6 y: I/Y is the mean I-cell lifetime in days; a cell lifetime ends if it either goes into
7 . mitosis, or cell death. Note that if one assumes that the probability for a cell to get
8 into mitosis is about the same as cell death then the mean cell lifetime can be
9 conveniently interpreted as time to mitosis (i.e., cell turnover time); thus, shorter
10 cell lifetime implies more frequent cell division. Note that the time to mitosis is a
11 random variable here, not a fixed constant as in the assumption made in the
12 Greenfield et al. (1984) model that has been used extensively by Cohen and Ellwein
13 (1988) to analyze experimental bladder cancer.
14
15 N: Number of (normal) target cells
16
17 G3.2. Practical Considerations
18 By statistical theory alone the E-M algorithm developed in this report provides an
19 elegant procedure which can be used to test hypotheses whether a particular parameter
20 is influenced by organics and carbon core individually or both together. For instance,
21 one could postulate that the parameter y (reciprocal of which represents mean cell
22 lifetime) is given by y(d,Dj) = YQ + y^d + y^^r and tnen proceed to test a null
23 hypothesis that yjj = 0, no effect of organics on cell lifetime. This temptation, however,
24 must be resisted because there would be too many parameters that must be estimated if
25 such statistical tests are to be performed. Therefore, rather than performing such a
26 statistical exercise, we proceed with a biologically plausible assumption that parameters q
27 and y depend only on lung burden of carbon core, C.
28 The duration of the Mauderly et al. study was about 940 days. To construct a dose-
29 response model with time-dependent lung burden, the time interval (0,940] is divided into
30 five subintervals; each subinterval spans 6 mo except for the last subinterval, which spans
31 from 730 (2 years) to 940 days. Corresponding to an ambient air concentration of diesel
32 emissions in mg/m , the deposition-retention model developed by Yu et al. is used to
33 calculate dosimetric (d, Dj), i « 1, 2, ..., 5, where organics dose, d, is not changing with
34 time because it reaches steady state quickly after exposure begins and Dj is the lung
35 burden of carbon core during the ith subinterval.
36
C-5 DRAFT-DO NOT QUOTE OR CITE
-------
1 The assumptions about dose-parameters relationship are given below.
2
3 1. The initiation rate associated with a lung burden {d, Dj, i = 1, 2, ..., 5} is given by
4 Mi(d,Dj) m Mo(i + a * d + b * Dj), for i = 1, 2,..., 5. This is the only parameter
5 that is assumed to depend on both d and D.
6
7 2. Probability of tumor formation from a malignant cell is assumed to be dependent
8 on lung burden D by q(Dj) = % + qjDj, i * 1, 2,..., 5. To simplify calculation, the
9 possibility that q is also dependent on organics d is not considered.
10
11 3. The cell lifetime parameter y is assumed related nonlinearly to lung burden D by
12 y(Dj) = y0 + YlLog(l + Dj), i = 1, 2, ..., 5.
13
14 To reduce the number of parameters that must be estimated from the Mauderty
15 data, some of the background parameters (MQ, qo» and YQ) f°r tne dose-response model
16 are estimated from the National Toxicology Program (NTP) historical control rate on
17 Fischer-344 rats (reconstructed from Portier et al., 1986). Giving these background
18 parameters, the dose-related parameters are then estimated by the E-M algorithm, which
19 is described in Appendix C-2. Using tumor response data from Mauderty et al. (1987)
20 and the corresponding dosimetric in Table C-l, the resultant parameter estimates for the
21 model are given in Table C-2. To have some appreciation about the implication of the
22 Mauderty et al. (1987) study, the estimated initiation and proliferation (for I-cells) rates
23 for the study are given in Table C-3. Although these values may not represent reality
24 (because they are not actual laboratory measurements), they could be used as a guidance
25 for future research planning. For instance, Table C-3 (along with a discussion about
26 Table C-7) suggests that a slight increase of proliferation rate could cause a drastic
27 increase on tumor incidence, but only if the initiation rate is high enough. This
28 conclusion seems to suggest that although the promotion effect of growth factors is
29 important for tumor induction, the initiation effect of carbon core and/or organics is also
30 essential.
31
C-6 DRAFT-DO NOT QUOTE OR CITE
-------
TABLE C-l. DOSIMETRIC (mg/cm2 lung surface) USE IN MODELING1
Exposure group d Dj
0.35
3.5
7.08
2.5E-6
3.6E-5
7.3E-5
6.23E-5
7.54E-4
1.98E-3
D2
8.75E-5
2.40E-3
5.49E-3
D3
8.97E-5
3.91E-3
8.56E-3
D4
9.02E-5
5.25E-3
1.12E-2
D5
9.02E-5
6.29E-3
1.44E-2
*d is organks; D4, i « 1,2,.... 5, are average lung burden of carbon core over five time intervals. These
values are calculated by Yu et aL retention model in Appendix D.
TABLE C-2. MAXIMUM LIKELIHOOD ESTIMATES
FOR MODEL PARAMETERS
Parameter Estimate
__1.Q33E-7
a 1.103E+4
b 3.214E+2
M2 7.907E-7
q0 1.035E-1
q! 5.332E-2
YO 1.662E-2
Yl 2.647
a 5.443E-1
Nb (given) 8.80E+7
'Background parameters ^ q^ and YO are estimated separately from NTP historical control data.
bTTie number of target cells N is assumed to be 10-fold of Type II cells in mice, which is given in
Kauffman
(1974). It is not essential for N to be given accurately because Np0 appears as a single term in the model;
the
estimated ^ will compensate for the under- or over-estimation of N.
C-7 DRAFT-DO NOT QUOTE OR CITE
-------
TABLE O3. RELATIVE MAGNITUDE OF INITIATION Ml (
-------
1 tumor mortality can only be calculated up to about 900 days because after 900 days
2 tumors are no longer discovered by natural mortality only; in fact, the majority of tumors
3 are discovered by sacrifice.
4
5
6 C5. RISK PREDICTIONS UNDER VARIOUS EXPOSURE
7 SCENARIOS
8 For comparison, excess lifetime risks (see Tables C-5 and C-6) due to various
9 exposure scenarios are calculated by the alternative model and the linearized multistage
10 (LMS) model. Both point (maximum likelihood estimate) and 95% upper bound
11 estimates are provided for the alternative model, whereas only upper bound estimate is
12 provided for the LMS model because its linear component (which is notoriously unstable)
13 is estimated to be 0. The 95% upper bound for the alternative model is calculated by
14 the same approach as for the LMS model; (i.e., by increasing parameters a and b until
15 the log-likelihood exceeds a critical value). To extrapolate from animal-based risk
16 estimates to human, two assumptions are made: (1) lung burden in terms of Mg/cm of
17 lung surface is equally potent between animals and humans, and (2) 6 mo of animal life
18 is equivalent to 18 years of human life. The latter assumption is necessary because life-
19 span must be divided into five subintervals to account for different parameter values.
20
C-9 DRAFT-DO NOT QUOTE OR CITE
-------
TABLE C-4. COMPARISON OF OBSERVED TUMOR MORTALITY RATE AND
PREDICTED PROBABILITY OF CANCER OCCURRENCE BY TIME (t) WHEN A
(MALIGNANT) TUMOR WAS OBSERVED IN RATS
Exposure Time Tumor Observed Tumor Mortality Predicted Tumor Rate
(mg/m ) Observed (days) Rate (95% C I.)
by Time CO
Control 538
551
0.35 710
863
3.50 891
895
7.08 646
672
701
729
742
798
810
839
840
847
856
859
883
895
0.0051
0.010
0.007
0.025
0.016
0.036
0.006
0.013
0.021
0.027
0.039
0.052
0.066
0.081
0.096
0.112
0.129
0.146
0.168
0.191
(0, 0.015)
(0, 0.028)
(0, 0.022)
(0, 0.063)
(0, 0.126)
(0, 0.052)
(0, 0.019)
(0, 0.037)
(0, 0.041)
(0, 0.059)
(0.004, 0.075)
(0.009, 0.095)
(0.016, 0.115)
(0.023, 0.138)
(0.032, 0.161)
(0.041, 0.183)
(0.052, 0.207)
(0.063, 0.229)
(0.077, 0.259)
(0.091, 0.291)
0.002
0.003
0.005
0.008
0.039
0.040
0.039
0.046
0.054
0.064
0.069
0.097
0.104
0.123
0.123
0.128
0.135
0.137
0.156
0.166
Calculated by the life table procedure. Note that observed values are mortality, which are expected to be
smaller than the (predicted tumor) incidence. This expectation is particularly true at early stage of tumor
development whet a tunor was small.
1
2 Table C-5 compares predicted risks for humans due to continuous exposure
3 (24 h/day) calculated by alternative and LMS models. It is interesting to see that risk
4 calculations under various exposure concentrations are very similar using the two
C-10 DRAFT-DO NOT QUOTE OR CITE
-------
1 different models. Table C-6 gives excess risks due to exposure to 2.6 /ig/m of diesel
2 emissions, 16 h/day, 7 days/week; and 15 jig/m , 8 h/day, 5 daysAveek. The
3 concentration 2.6 ng/m was reported by EPA's Office of Mobile Sources to be the
4 annual mean exposure of the U.S. population to diesel paniculate matter in 1986 and is
5 only slightly higher than the most recent estimate of 2.03 ngfm in an EPA draft
6 document (Motor Vehicle-Related Air Toxic Study, April, 1993); the concentration 15
7 Mg/m was reported to be the paniculate exposure for workers on urban freeways in an
8 EPA report by Carey (Air Toxics Emissions from Motor Vehicles, 1987, EPA-AA-TSS-
9 PA-86-5). For the general population exposed to an ambient air concentration of 2.6
10 /jg/m , the risk to normal (i.e., persons with normal respiratory functions) and smokers of
11 20 pack-years (as defined by Bohning et al., 1982) are provided. According to Bohning
12 et al. the retention half-life for insoluble particle increases from 296 days for persons with
13 normal respiratory function to 519 days for persons with a smoking history of 20 pack-
14 years. This information is used to reduce the alveolar clearance rate in the dosimetric
15 calculations using the same retention model that is also used to calculate dosimetric in
16 Table C-l.
17 The excess lifetime risks in Tables C-5 and C-6 are calculated by actuarial life-table
18 approach using the survival probability of the NTP control animals provided in Portier
19 et al. (1986). Conceptually, this approach can be viewed as a weighted average of the
20 probability of cancer occurrence over entire lifetime, weighted by survival probability.
21 This approach is more appropriate than the one used previously in the draft report in
22 which probability of cancer occurrence at a preselected time (730 days) is used to
23 represent the lifetime risk; it is more appropriate because tumors here occur very late in
24 life.
25
C-ll DRAFT-DO NOT QUOTE OR CITE
-------
TABLE C-5. COMPARISON OF EXCESS RISK FOR HUMANS DUE TO
CONTINUOUS EXPOSURE OF VARIOUS CONCENTRATIONS OF DIESEL
EXHAUST EMISSIONS UNDER TWO DIFFERENT MODELS
Alternative Model
Exposure Concentration MLE 95% u.b. LMS Model
(Mg/m ) ^
0.01 7.68E-8 1.35E-7 1.71E-7
0.1 8.12E-7 1.41E-6 1.72E-6
1.0 (unit risk) 8.16E-6 1.65E-5 1.74E-5
100 5.58E-4 9.63E-4 1.74E-4
1.000 2.60E-2 4.22E-2 3.33E-2
"MLE: maximum likelihood estimate; 95% u.b.: 95% upper bound estimate. These are calculated using
the
alternative dose-response model
T-MS: calculated by linearized multistage model (slope « 9.04 per mg/cm of lung surface), using carbon
core as dosimetric Only malignant tumors are used in the calculations.
TABLE C-fc EXCESS LIFETIME RISK FOR HUMANS DUE TO EXPOSURE TO
DIESEL EXHAUST EMISSIONS, UNDER VARIOUS EXPOSURE SCENARIOS
Exposure Pattern
2.6 jig/m , 16 h/day, 7 days/week
Normal person
2.6 jig/m , 16 h/day, 7 days/week
20 pack-year smoker
15 Mg/m » 8 h/day, 5 days/week
Alternative Model
MLE 95% u.b.
1.41E-5 2.44E-5
2.32E-5 3.61E-5
3.12E,5 5.17E-5
LMSb
3.00E-5
5.38E-5
6.18E-5
*MLE: maximum likelihood estimate; 95% u.b.: 95% upper bound estimate. These are calculated using
the
alternative dose-response model
\MS: calculated by linearized multistage model using carbon core as dosimetric Only malignant tumors
are used in the calculations.
1 C.6. DUPLICATIONS OF THE ALTERNATIVE MODEL
C-12 DRAFT-DO NOT QUOTE OR CITE
-------
1 Before proceeding to discuss implications of the alternative model on risk
2 assessment, it should be noted that the parameters used in the model are estimated on
3 the basis of high exposure concentration cancer bioassay data, not on the basis of data
4 from laboratory measurements (e.g., mitotic rates for cells from normal and preneoplastic
5 . lesions measured over time), which usually can be obtained over lower range of exposure
6 concentrations. Therefore, uncertainty associated with low-dose extrapolation still
7 remains. For this reason, we will refrain from using the model to evaluate low-dose risk
8 estimations, but rather to evaluate the relative contribution of each biological component
9 (e.g., initiation by organic and carbon core, individually or jointly) in the model to cancer
10 induction.
11 On the basis of the constructed alternative dose-response model, some specific
12 inferences could be made from Table C-7 by changing parameter values of the original
13 model. Table C-7 provides a comparison of risks calculated with changed parameters,
14 assuming that animals are exposed to 7.08 mg/m of diesel exhaust emissions, 7 h/day, 5
15 days/week (which is identical to the exposure pattern of the highest exposed group in
16 Mauderly et al, 1987).
17
TABLE C-7. EFFECT OF CHANGING DOSE-DEPENDENT
INITIATION AND PROMOTIO^ PARAMETERS
5 days/week for life [Le, the highest exposure group in Mauderly et aL, 1967].)
Case
Number
1
2
3
4
5
6
7
8
Parameters Changed
None
(original model)
a = 0
b =: 0
a~0
b = 0
a = 2.813aa
Yj = l,378yj
Y! = 1.756yj
Y,-0
Risk at 938 Days
0.2067
0.0663
0.1616
0.1165
0.2007
0.3513
0.4537
0.0319
Ratio to Original Model
1.00
0.321
0.782
0.564
0.971
1.700
2.195
0.154
8a = 2.813a implies that a is increased by 2.813 times of its original value.
C-13 DRAFT-DO NOT QUOTE OR CITE
-------
1 The following observations can be made from Table C-7.
2
3 1. When there is no diesel induced initiation (Case 2), the risk is 32% of the original
4 model (i.eM the model without changing parameters), iiucontrast to 42% when
5 exposure concentration is reduced from 7.08 to 1 mg/m (not shown here).
6 Therefore, the role of diesel-induced initiation in cancer induction increases with
7 increasing exposure concentrations. This conclusion is intuitively obvious because
8 spontaneous induction of initiated cells play a bigger role in cancer induction when
9 concentration is lower. A practical implication of this observation is that reduction
10 of non-diesel-induced initiation (e.g., by smoking) could have greater proportion of
11 cancer risk reduction when diesel concentration is low than when the concentration
12 is high.
13
14 2. Cases 3 and 4 indicate that initiation by either carbon core, or organics contributes
15 significantly to tumor incidence.
16
17 3. Case 5, along with observation Number 2 above, suggests that although diesel-
18 induced I-cells play an important role in cancer induction, the role of initiation,
19 however, could be assumed by either organics or carbon core alone by increasing
20 their respective proportions. One implication is that, although existence of I-cells
21 are important for tumor induction, these I-cells could be induced by any agent that
22 initiates (e.g., smoking).
23
24 4. Cases 6 to 8 suggest that a small change of proliferation parameter y could have a
25 disproportionate change of cancer risk. Because this parameter is assumed a
26 function of carbon core dose, lung burden overloading has a significant effect on
27 cancer incidence. In the absence of better information, it is assumed in this report
28 that carbon core continues to have effect at low doses.
29
30 These four observations together suggest that while effect of growth factors (which
31 may increase value of y) by particle overloading is important, the initiation effect of
32 carbon core and/or organics is also essential. Although this conclusion is only tentative
33 because the model parameters are estimated on the basis of high concentration bioassay
34 data, they do suggest the importance of studying the role of carbon core and organics on
35 initiation and promotion at low- versus high-exposure concentrations. Does the relative
36 initiation potential between organics and carbon core differ at high and low
37 concentrations? Along with the results in Table C-6, it also suggests that a subcohort of
38 workers who were smokers and exposed to high concentrations of diesel exhaust for a
39 long duration would be expected to have higher lung cancer mortality.
C-14 DRAFT-DO NOT QUOTE OR CITE
-------
1 It is interesting to observe from Table C-8 that, under the same exposure
2 conditions, risk is greater when exposure begins later in life. This model-based
3 conclusion is due to the fact that older animals have more spontaneously (including non-
4 diesel) induced initiated cells that have potential to be proliferated, converted to
5 malignant cell, and then progressing to cancer. (Note that the above observation would
6 not contradict any observation that might show that younger animals are more sensitive
7 to diesel exposure than older animals if a treatment induces more initiated cells in the
8 younger animals). Assuming that the above model-generated hypothesis is realistic, an
9 important implication is that initiation by nondiesel agents should be considered when
10 assessing risk to humans due to exposure to diesel emissions.
11
TABLE C-8. EXCESS CANCER RISK TO RATS 200 AND 300 DAIS AFTER
TERMINATION OF 6-MO EXPOSURE TO D - 730E-5 mg/cni OF
ORGANICS, AND D - 1.89E-3 mg/cm OF CARBON CORE*
Days After Exposure Terminated
Exposure Period (mo) 200 300
1.48E-3 1.96E-3
6 to 12 4.01E-3 5.25E-3
12 to 18 _ 8.32E-3 _ 1.05E-2 _
*The lung burden is assumed to be zero over unexposed periods. It may not be a realistic assumption
because
the lung burden is expected to linger over the following periods after exposure terminated; however, the
assumed exposure condition serves the purpose better here.
1 C7. CONCLUSIONS AND SUMMARY
2 1. The risk predictions by both alternative and LMS models are comparable over a
3 range of exposure concentrations that is of practical interest. However, this
4 conclusion is valid only under the assumption that the effect of carbon core on each
5 biological component (e.g., initiation) in the model continues to exist at low doses
6 (see further discussions about uncertainties below). Based on the Mauderty et al.
7 (1987) study, the risks associated with continuous exposure to 1 jig/m of diesel
8 emissions calculated by two different models are summarized below:
9
C-15 DRAFT-DO NOT QUOTE OR CITE
-------
Alternative Model
Lung Tumor Data Used
Mlignant tumors
Total tumors
MLE
8.16E-6
NA
95% u.b.
1.65E-5
NA
LMS Model
95% u.b.
1.74E-5
3.44E-5
(taken from Chapter 11)
1 2. The model suggests that populations with higher expected background rate (e.g.,
2 smokers) may be subjected to higher lung cancer risk than the populations with
3 lower background rate. It is noted that U.S. females have about the same
4 background lung cancer rates as the Fischer-344 rats (about 1 to 2%), whereas U.S.
5 males have a background rate of 6%. However, because most of lung cancers are
6 smokers, the risk to nonsmokers (males or females) should be about the same. The
7 use of the unit risk estimate provided in Chapter 11 may somewhat underestimate
8 risk to smokers (or other respiratory-impaired persons) unless adjustment on lung
9 burden is made. Table C-6 provides an example of such adjustment.
10
11
12 C8. DISCUSSIONS ABOUT UNCERTAINTIES OF RISK ESTIMATES
13 Although, it is interesting to note that risk estimates by the LMS model are
14 comparable to those calculated by the alternative model, there are uncertainties about
15 low-dose extrapolation by the alternative (as well as by the LMS) model: first, the model
16 parameters are estimated statistically, not measured in the laboratory; and second, the
17 model parameters are estimated on the basis of high-exposure data, the relationship
18 between a parameter and exposure below the exposure range remains unknown, and the
19 dose-parameter relationship used in the model may not be adequate for low-dose
20 extrapolation. For instance, it is assumed that initiation rate is linearly related to doses
21 of carbon core. Such an assumption needs be evaluated. The risk at low doses would be
22 overestimated in this report if the relationship between initiation rate and carbon core is
23 sublinear (concave upward). The sublinear assumption would be reasonable if there is
24 no effect of initiation by carbon core dose (D) at low concentration. On the other hand,
25 the risk would be underestimated if the relationship is supralinear. Therefore, it is
26 important to evaluate how increase of diesel-exposure concentration affects initiation rate
C-16 DRAFT-DO NOT QUOTE OR CITE
-------
1 over low-exposure concentrations. Similarly, it is important to know the relationship
2 between dose of carbon core and cell (I-cells in particular) proliferation at low
3 concentration.
4 Another aspect of uncertainty is the use of lung burdens (organics and carbon core)
5 calculated by mathematical model, rather than actually measured. However, the impact
6 of this uncertainty with regard to the conclusions reached in this report is not expected to
7 be significantly altered even if the model-based dosimetrics are not accurate because the
8 relative patterns of lung burdens between high- and low-exposure concentrations, and
9 between animals and humans should be about the same. Although there is some
10 observed total lung burden, these data are not used because of the following reasons.
11
12 1. The observed data are not separated by organics and carbon core.
13
14 2. There are no human data—these data are needed to predict risk in humans.
15
16 3. The observed data do not go beyond 730 days.
17
18 4. The desire is to be consistent with Chapter 11 so that results can be compared.
19
20
21 C9. DISCUSSIONS ABOUT FUTURE RESEARCH NEEDS
22 The single most important use of a biologically based dose-response model in the
23 cancer risk assessment is to reduce uncertainty of low-dose extrapolation when the
24 mechanism for tumor response observed at high doses differs drastically from the low
25 doses. However, this report can focus only on the use of the model to guide future
26 research rather than to actually reduce uncertainty of risk estimate because of our
27 inability to obtain biological parameters in the model. If a chemical is known to induce
28 disproportionately larger cell proliferation (in normal, initiated, and/or malignant cells) at
29 high doses than at low doses, then a model that reflects this fact would be useful. With
30 this in mind, our effort should be to identify "components" of carcinogenesis (e.g.,
31 increase of mitotic rate) that are disproportionately more affected at high doses than at
32 low doses and to develop models that incorporate those high-dose effects. For the diesel
33 risk assessment, the "components" that require further study include effects of organics
C-17 DRAFT-DO NOT QUOTE OR CITE
-------
1 and carbon core, individually or jointly, on initiation, proliferation, conversion, and
2 progression steps of carcinogenesis. In order to use biologically based models of
3 carcinogenesis in risk assessments, one needs to know the relationship between
4 parameter values in a model and exposure (or dose). Ideally, some of these parameters,
5 if not all, should be measured directly in the laboratory, or indirectly estimating from
6 neoplastic and preneoplastic lesions (e.g., number of foci, adenomas, and tumors in a
7 lung).
8 Cell proliferation is an increase in the cell population of different stages: normal,
9 initiated, or malignant cells. Enhanced cell proliferation of normal target cells may itself
10 increase the frequency of mutations, either by inducing error in replication or by
11 converting DNA adducts to mutations before DNA repair can occur. The model implies
12 that tumor incidence is linearly proportional to initiation rate. On the other hand,
13 enhanced cell proliferation of initiated cells could lead to more than linear increase of
14 tumor incidence. Therefore, proliferation of I-cells has a greater impact on tumor
15 incidence than proliferation of normal cells. However, this does not mean that initiation
16 potential of compounds (organics or carbon core) is not important. As discussed
17 previously, it is important to determine the ability of these compounds to initiate at low
18 versus high doses; this has a significant implication for low-dose extrapolation. From the
19 viewpoint of mathematical modeling, cell proliferation is the result of a decrease of cell
20 death rate and/or an increase of mitotic rate, regardless of underlying biological
21 mechanisms. Therefore, it is logical to construct a model (as is done here) with a
22 proliferation component involving cell death and mitosis, and important to obtain data at
23 the cellular level even if biological mechanism at the molecular level is not yet known. If
24 a more precise mechanism is known and the quantitative data are available, then the
25 proliferation component of the model can be improved by incorporating the available
26 biological information. Most of the two-stage models consider a single malignant cell to
27 be equivalent to a tumor. If a compound is known to affect the cell proliferation of
28 tumor cell population, a model that incorporates tumor progression should be used. For
29 the diesel modeling, we assume that particles could enhance the proliferation of
30 malignant cells. This assumption needs to be verified. Another model-generated
31 hypothesis is that persons with higher number of initiated cells are subjected to higher
C-18 DRAFT-DO NOT QUOTE OR CITE
-------
1 lung cancer risk when exposed to diesel emissions. (A person could have a higher
2 number of initiated cells due to exposure to diesel and/or nondiesel agents, or simply by
3 acquiring more spontaneously induced initiated cells through aging).
4 In summary, information that is necessary to construct a biologically based dose-
5 response model includes (1) identifying roles that are played by organics and carbon core
6 (individually or jointly) with respect to initiation, proliferation, conversion, and
7 progression, at low versus high doses; (2) quantitative measurements of cellular dynamics
8 (e.g., mitotic rate) for cells at different stages and exposure concentrations; and (3)
9 relationship between parameters and exposure or dose. Because many biological
10 parameters are expected to be age-dependent, they should be measured over different
11 time points. Furthermore, frequency and size of preneoplastic foci, nodules, and tumors
12 could also provide useful information toward improving risk assessment. Some of these
13 data may be obtained by initiation-promotion type of study.
14
15
C-19 DRAFT-DO NOT QUOTE OR CITE
-------
l CIO. REFERENCES
2
3 Bohning D., Atkins, H., and Conn, S. Long-term panicle clearance in man: normal and impaired. Ann.
4 Occup. Hyg. 26:259-271.
6 Chen, C, and Farland, W. Incorporating Cell Proliferation in Quantitative Cancer Risk Assessment:
7 Approach, Issues, and Uncertainties. In: Chemically Induced Cell Proliferation: Implications for
8 Risk Assessment, B. Butterworth, T. Slaga, W. Farland, and M. McCain, eds.; pp. 481-499
9 (Wiley-Liss, Inc. New York, 1991).
10
11 Cohen S. and EUwein L. Cell growth dynamics in long-term bladder carcinogenesis. ToricoL Letters,
12 43:151-173, 1988.
13
14 Dempster, A., Laird, N., aad Rubin, D. Maximum likelihood from incomplete data via the EM algorithm.
15 Roy. Statist Soc, B 39:1-38, 1977.
16
17 Greenfield, R., Ellwein, L, and Cohen, S. A general probabilistic model of carcinogenic analysis
18 experimental bladder cancer, Carcinogenesis 5:437-445,1984.
19
20 Kauffman S. Kinetics of alveolar epithelial hyperplasia in lungs of mice exposed to urethane I. quantitative
21 analysis of cell populations, Laboratory Investigations, 30:170-175, 1974.
22
23 S. Moolgavkar and D. Venzon, Two-Event Models for Carcinogenesis: Incidence Curve for Childhood and
24 Adult Tumors/Math. Biosciences 47, 55-77 (1979).
26 S. Moolgavkar and A Knudsoa, 'Mutation and Cancer: A Model for Human Carcinogenesis,' Journal of
27 the National Cancer Institute 66, 1037-1052 (1981).
28
29 Ponier C, Hedges, J., and Hoel, D. Age-specific models of mortality and tumor onset for historical control
30 animals in the National Toxicology Program's carcinogenicity experiments. Cancer Research
31 46:4372-4378, 1986.
32
33 W. Tan and C Chen, A nonhomogeneous stochastic model of carcinogenesis for assessing risk of
34 environmental agents. To appear in: Mathematical Population Dynamics, Proceedings of the Third
35 International Conference, Pau, France, 1992.
36
C-20 DRAFT-DO NOT QUOTE OR CITE
-------
i APPENDIX C-l
2 E-M ALGORITHM
3
4 The E-M algorithm is derived below. It will be used to calculate maximum
5 likelihood estimate of parameters of the alternative model. Data used for the E-M
6 algorithm is taken from Mauderly et al. (1987), which includes time when an animal died
7 (natural mortality or sacrifice) with or without (malignant) tumors. The computer
8 program for the E-M calculations was developed by Mr. Daliang Chang of the Computer
9 Science Corporation under an EPA contract. The theory of E-M algorithm can be found
10 in Dempster et al. (1977).
11 Assume that the distinct times when animals died by either natural mortality or
12 sacrifice are tj < t2 < ...< t^ The observations can be classified as follows:
13
14 aix(i): observed number of natural deaths without tumor at time tj in the
15 treatment group x (There are four groups for diesel data [i.e., x = 1, 2,
16 3, 4.]),
17
18 a2x('): observed number of natural deaths with tumor at time tj in the
19 treatment group x,
20
21 bix(0: series sacrifice at time tj without tumor in the treatment group x,
22
23 b2x(i): series sacrifice at time t, with tumors in the treatment group x.
24
25
26 Let Td represent the time an animal died and T the time a tumor developed.
27
28 ax(i) = Pr{Td * tj|Td > tj, T > tj, x} (conditional probability of death without
29 tumor)
30
31 0x(i|u) - Pr{Td = tj|Td > tj, Te(tu j, tj, x} (related to deaths with tumors)
32
••% ^
jj
34 Define,
35
36
C-21 DRAFT-DO NOT QUOTE OR CITE
-------
n [i - p,aiu)].
S,(t) - PriT * t|x) - exp[-Jh(x)dx].
1
2 The function Sx(t) is the probability of tumor free by time t. The exact form of the
3 hazard function h(x) and Sx(t) are given in the next section.
4
5 Let
6
7 a2x('lu) * number of natural death at tj with tumor developed during (t^j, tj, in
8 the treatment group x, u < i,
9
10 b^Olu) « number of sacrifice at tj with tumor developed during (ty.j, tj, in the
11 treatment group x, u < i,
12
13 Then
14
15 Let
16
J-l
17
18 and
19
20
21
22 where
24
C-22 DRAFT-EX) NOT QUOTE OR CITE
-------
1 Given a^OO, {a2x('lu)» u = lj •"» ^' *s an 0 ~ l)-d>roension multinomial with
2 parameter {a^O), PXOIU)> u = 1, ..., i}.
3
4 Thus, £[8^0 1 u) | a^Ci)] = a2x(i)Px(i | u).
5
6 Similarly, {b^Olu), u = 1, ..., i}, is an (i - l)-dimension multinormial with parameters
7 {b2x(j)> QX(>|U). u = 1. -. »>. and
8
9 Efb^Ci | u) | b^i)] = b2x(i)Qx(i | u).
10
11 It can be shown that the likelihood function is proportional to
z M
12
13 where
• • i
14
15 Let
KI^) • E t«uO) * buO) * mx(j)], and
J-i
16
17 Let
6 " (Up Vy Yp •••)
18
19 be a vector of parameters in function S;
a, « [as(l), a ,(2) ..... a,(m)]f and
20
C-23 DRAFT-DO NOT QUOTE OR CITE
-------
1 be vectors of parameters related to conditional probabilities of death with and without
2 tumors. These parameters, along with those in 8X will be estimated by the E-M
3 algorithm described below.
4
5 The M -step:
6 Given initial values a2x(i | u) and b2x(i | u), estimate
1. o> - au(i)/Ru(i)
2. p^^ifcCMfRfcOW, and
3. obtain 6X by m»*immn% the log of L.
8
9 TheE-Step:
10 Given the estimated values on o^i), 0x(i), and ex from the M-step, compute Px(i|u) and
1 1 Qx(i | u), and obtain estimates of a2x(i | u) and b2x(i | u) by
12
, and
13
14 With the estimated values of a2x(i | u) and b2x(i | u) available from the E-step, go
15 back to the M-step. Repeat the same process until estimates are stabilized.
16
17
18
C-24 DRAFT-DO NOT QUOTE OR CITE
-------
i APPENDIX C-2
2 A TUMOR GROWTH MODEL
3
4 The tumor growth model with piece-wise constant parameters is taken from Tan
5 and Chen (1992), which is an extension of a stochastic model developed by Chen and
6 Farland (1991). This model has a similar biological motivation as the two-stage model
7 proposed by Greenfield et al. (1984), which has been used by Cohen and Ellwein (1988)
8 to analyze bladder tumors. However, the two models differ from each other with respect
9 to their mathematical formulations; the one adopted in this report is a stochastic model,
10 whereas the other is a deterministic model and does not allow for parameters estimation
11 because the model does not have complete mathematical expression.
12 Although its most general form will not be used here because of the lack of data, it
13 is worthwhile to note that the stochastic model by Chen and Farland (1991) has two
14 desirable features: (1) it allows for any cell growth distributions (e.g., Gompertz), rather
15 than limiting only to the exponential distribution as in other existing models; and (2) it
16 incorporates the birth and death of tumor cells, rather than assuming that a tumor is
17 born once a single tumor cell occurs, an assumption made by the MVK model (a model
18 proposed by Moolgavkar et al., 1979, 1981). Therefore, if information on cell lifetime
19 distribution, and the progression phase of tumor development is available, a reasonably
20 realistic model can be constructed.
21 For completeness of the report a brief description of the model will be presented
22 here. The following notations are needed for the model:
23
24 N(t): number of normal (target) cells at time t,
25
26 MI' initiation rate, and
27
28 f(t): the probability density function for the lifetime of an initiated cell (I-cell).
29
30 For an I-cell, at the end of its lifetime it either divides (mitosis) or dies
31 (programmed or nonprogrammed death). If it enters into mitosis, it either divides into
32 two I-cells with probability a, or divides into one I-cell and one malignant cell (M-cell)
C-25 DRAFT-DO NOT QUOTE OR CITE
-------
1 with probability ^2- N°te that, at the end of a cell's lifetime, the probability for the cell
2 to die is /? = 1 - a - ^2- A similar setup (i.e., to allow for any cell lifetime distribution)
3 can be made for an M-cell. However, we will confine ourselves to a simpler version
4 assuming that an M-cell lifetime follows an exponential distribution. Thus, we can simply
5 assume that an M-cell follows a simple birth-death process; it can either divide into two
6 M-cells with a rate am or die with a rate 0m.
7 When parameters are constant over time (ages), the hazard function is given by
t
h(t) = M,M2fN(s)m(t - s) ds
8 where
& - y,)2exp[A(t)a(y2 - y,)]
((1 - y,) + (y2 - l)exp[A(t)a(y2 - y,)])2'
9
2
10 where yj < y2 are two real roots of oy - (a + fl + M2
-------
where
k
nm& ~ li i) s *» when J = k
J J j '
2
3 and
^(t). <>»-'
x1 - vij) * (y? - l)«piA,(t)o.(y, - y^)])2
4
2
5 where yj: < y2j are two real roots of a3 - (a: + 0; + M2jclj)y + 0j = 0; a; + 0: +
7 When exponential distribution (i.e., Aj(t) = y:t and q: = 1 are assumed, the model
8 is equivalent to the MVK model with piece-wise constant parameters. A special case
9 that may be more appropriate than the exponential distribution is when the Gompertz
10 distribution is assumed (i.e., when Aj(t) = {1 - exp[-Yjt]}/Yj).
11 For the diesel alternative model, the total time is divided into five (i.e., k » 5)
12 subintervals. It is shown in Tan and Chen (1992) that, under the assumption of
13 exponential cell lifetime distribution, the tumor free distribution function, Sx(t), can be
14 written as
15
k j-i
S(t) « exp(-£ [Ay(tH, «p +
j-i i«i
16
17 where S: = t: if j < k and s: « t if j = k, and
18
19
20
21
C-27 DRAFT-DO NOT QUOTE OR CITE
-------
w, + z, + (w, -
los[wi
w, * z, + (w, -
1
2 where,
3
4 Wj - [(a + f + M2q)2 -
5 Zj
6 Ajj(s,t) = yj(t - s) if both s and t are in the same closed subinterval [tj.j, tj] and
7
j-i
8
if seL;, teLj with t. < t.
9
10
11
C-28 DRAFT-DO NOT QUOTE OR CITE
-------
APPENDIX D
MODELS FOR CALCULATING LUNG BURDENS
December 1994 DRAFT-DO NOT QUOTE OR CITE
-------
APPENDIX D
MODELS FOR
CALCULATING LUNG BURDENS
D.I. INTRODUCTION
As discussed in Chapter 4 the lung burden of diesel exhaust particles (DEPs) during
exposure is determined by both the amount and site of particle deposition in the lung and,
subsequently, by rates of translocation and clearance from the deposition sites. Mathematical
models have often been used to complement experimental studies in estimating the lung
burdens of inhaled particles in different species under different exposure conditions. This
section presents a mathematical model that simulates the deposition and clearance of DEPs in
the lungs of rats and humans.
Diesel particles are aggregates formed from primary spheres of 15-30 nm in diameter.
The aggregates are irregularly shaped and range in size from a few molecular diameters to
tens of microns. The mass median aerodynamic diameter (MMAD) of the aggregates is
approximately 0.2 um. The primary sphere consists of a carbonaceous core (soot) on which
numerous kinds of organic compounds are adsorbed. The organics normally account for 10%
to 30% of the particle mass. However, the exact size distribution of DEPs and the specific
composition of the adsorbed organics depend upon many factors, including engine design,
fuels used, engine operating conditions, and the thermodynamic process of exhaust. The
physical and chemical characteristics of DEPs have been reviewed extensively by Amann and
Siegla (1982) and Schuetzle (1983).
Four mechanisms deposit diesel particles within the respiratory tract during exposure:
impaction, sedimentation, interception, and diffusion. The contribution from each mechanism
to deposition, however, depends upon lung structure and size, the breathing condition of the
subject, and particle size distribution. Under normal breathing conditions, diffusion is found
to be the most dominant mechanism. The other three mechanisms play only a minor role.
Once DEPs are deposited in the respiratory tract, both the carbonaceous cores and the
adsorbed organics of the particles will be removed from the deposition sites as described in
D-l DRAFT-DO NOT QUOTE OR CITE
-------
Chapter 4. There are two mechanisms which facilitate this removal: (a) mechanical clearance,
provided by mucociliary transport in the ciliated conducting airways as well as macrophage
phagocytosis and migration in the nonciliated airways, and (b) clearance by dissolution. Since
the carbonaceous soot of DEPs is insoluble, it is removed from the lung primarily by mechanical
clearance, whereas the adsorbed organics are removed principally by dissolution.
D.2. PARTICLE MODEL
To develop a mathematical model which simulates the deposition and clearance of
DEPs in the lung, an appropriate particle model characterizing a diesel particle must first be
introduced. For the deposition study, we employed an equivalent sphere model for the diesel
particle developed by Yu and Xu (1987) to simulate the dynamics and deposition of DEPs in
the respiratory tract by various deposition mechanisms. For the clearance study, we assume
that a diesel particle is composed of three different material components according to their
characteristic clearance rates: (1) a carbonaceous core of approximately 80 percent of the
particle mass, (2) absorbed organics of about 10 percent of particle mass, and that are slowly
cleared from the lung (3) adsorbed organics quickly cleared from the lung accounting for the
remaining 10 percent of particle mass. The presence of two discrete organic phases in the
particle model is suggested by observations that the removal of particle-associated organics
from the lung exhibits a biphasic clearance curve (Sun et al., 1984; Bond et al., 1986) as dis-
cussed in Chapter 4. This curve represents two major kinetic clearance phenomena: a fast
phase organic washout with a half-time of a few hours and a slow phase with a half-time that
is a few hundred times longer. The detailed components involved in each phase of the clear-
ance are not known. It is possible that the fast phase consists of organics which are leached
out primarily by diffusion mechanisms while the slow phase might include any or all of the
following components: (a) organics which are "loosened" before they are released, (b)
organics which have become intercalated in the carbon core and where release is thus
impeded, (c) organics which are associated for longer periods of time due to hydrophobic
interaction with other organic phase materials, (d) organics which have been ingested by mac-
rophages and as a result effectively remain in the lung for a longer period of time due to
metabolism by the macrophage; metabolites formed may interact with other cellular compo-
D-2 DRAFT-DO NOT QUOTE OR CITE
-------
nents, and (e) organics which have directly acted on cellular components such as the forma-
tion of covalent bonds with DNA and other biological macromolecules to form adducts.
The above distinction of the organic components is largely mechanistic and it does not
specifically imply the actual component nature of the organics adsorbed on the carbonaceous
core; however, the distinction is made to account for the biphasic clearance of DEPs.
However, this distinction is necessary in appreciating the dual phase nature of DEPs. For
aerosols made of pure organics such as benzo(a)pyrene (BaP) and nitropyrene (NP) in the
same size range of DEPs, Sun et al. (1984) and Bond et al. (1986) observed a nearly
monophasic clearance curve. This might be explained by the absence of intercalative
phenomena (a) and of hydrophobic interaction imposed by a heterogeneous mixture of
organics (b). The measurement of a pure organic might also neglect that quantity which has
become intracellular (c) or covalently bound (d).
D.3. COMPARTMENTAL LUNG MODEL
To study the transport and removal of DEPs from the lungs, we used a compartmental
model consisting of four anatomical compartments: the nasopharyngeal or head (H), tracheo-
bronchial (T), alveolar (A), and lung associated lymph node (L) compartments as shown in
Figure D-l. In addition, we used two outside compartments B and G representing, respec-
tively, the blood and gastrointestinal (GI) tract. The alveolar compartment in the model is
obviously the most important compartment for long-term retention studies. However, for
short-term consideration, retentions in other lung compartments may also be significant. The
presence of these lung compartments and the two outside compartments hi the model therefore
provides a complete description of all clearance processes involved.
In Figure D-l, r %, rty, and r ^ are, respectively, the mass deposition rates of DEP
material component i (i=l (core), 2 (slowly cleared organics), and 3 (rapidly cleared
organics)) in the head, tracheobronchial and alveolar compartments; and
D-3 DRAFT-DO NOT QUOTE OR CITE
-------
r— Ji 0)
, (I)
, ATB
B
^ AAB
\
_ ALB
rH
o m
HHG ,
*~ i
r-r0 G
<> . (i)
T ATG .
1 *"|
n ,|N (I)1" '
\i/ *•
1 I A
*v AT" f—^
,,c
Figure D-l
Compartmental model of DEP retention.
D-4 DRAFT-DO NOT QUOTE OR CITE
-------
represents the transport rate of material component i from any compartment X to any compartment Y.
Let the mass fraction of material component i of a diesel particle be //'. Then
$-firH.
amH = r(0 _ ^(0 m(i) _ x(0 (0 (D-7)
Tracheobronchial (T)
rfm(0
T _ JO . j(0,_(0 i<0_,(0 i<0m(0 (D-8)
—— - rT + t^MmA - ^•mmT - ^TBmT ,
D-5 DRAFT-DO NOT QUOTE OR CITE
-------
Alveolar (A)
^-(0 ^ n
,(0 i«M(0 (D-9)
df '^
Lymph Nodes (L)
dm®
Equation D-9 may also be written as
— - r(f) X(0/M(0 ^D"11^
rf/ " A ~ A A '
j(0 _ ,(0 ,(0 j(0 (D-12)
where *» " ^r A^ A^ '
is the total clearance rate of material component i from the alveolar compartment. In equa-
tions D-7 to D-10, we have assumed vanishing material concentration in the blood compart-
ment to calculate diffusion transport.
The total mass of the particle-associated organics in compartment X is the sum of m
and m ^ the total mass of DEPs in compartment X is equal to
m -
The lung burdens of diesel soot (core) and organics are defined, respectively, as
-2.
and
D-6 DRAFT-DO NOT QUOTE OR CITE
-------
Because the clearance of diesel soot from compartment T is much faster than from
compartment A, m ty ^ m ty a short time after exposure, equation D-14 leads to
Solution to equations D-7 to D-10 can be obtained once all the transport rates Xy
are known. When X^ are constant, which is the case of linear kinetics, equations D-7 to
E-10 will have a solution that increases with time at the beginning of exposure but eventually
saturates and reaches a steady-state value. This is the classical retention model developed by
the International Eommission of Radiological Protection (ICRP, 1979). However, as
discussed in Chapter 4, data have shown that when rats are exposed to DEPs at high
concentration for a prolonged period, the diesel soot accumulates in various peribronchial and
subpleural regions in the lung and the long-termed clearance is impaired. This is the so-called
overload effect, observed also for other insoluble particles. The overload effect cannot be
predicted by the classical ICRP model. Soderholm (1981) and Strom et al. (1987, 1988) have
proposed a model to simulate this effect by adding a separate sequestrum compartment in the
alveolar region. In the present approach, a single compartment for the alveolar region of the
lung is used and the overload effect is accounted for by a set of variable transport rates \^T ,
\($L, and X^ which are functions of mA. The transport rates X^ and X^L in equations D-7
to E-10 can be determined directly from experimental data on lung and lymph node burdens,
and Xr and XB from equation D-12.
D.4. SOLUTIONS TO KINETIC EQUATIONS
Equation D-ll is a nonlinear differential equation of m $ with known function o
For diesel soot, this equation becomes
Because clearance of the particle-associated organics is much faster than diesel soot, m and
m^3J constitute only a very small fraction of the total particle mass (less than one percent) after
a long exposure and we may consider X^ as a function of mty alone. Equation D-17 is
then reduced to a differential equation with m^J the only dependent variable.
D-7 DRAFT-DO NOT QUOTE OR CITE
-------
The general solution to equation D-17 for constant rflj at any time, t, can be obtained by
the separation of variables to give
I.
rA
m
= t . (D-18)
If r'J is an arbitrary function of t, equation D-17 needs to be solved numerically such as
by a Runge-Kutta method. Once mflj is found, the other kinetic equations D-7 to D-10 for
both diesel soot and the particle-associated organics can be solved readily, since they are
linear equations. The solutions to these equations for constant /ft , rty, and rty are given
below:
Head (H)
vhere X - c + X
Tracheobronchial (T)
«}? = exp (-X? O |0' ( r(/} * X?r m® ) exp (X®. / ) A * < (D-21)
X?
Lymph Nodes (L)
« 0 f 0' ^$WX«) dr * mJJ) (D-23)
In equations D-19 to D-23, m ty0 represents the value of m at t = 0.
In the sections to follow, the methods of determining r$ , /$, and r% , or (DF)H, (DF)T,
and (DFA /^ » rrD?» and ^5 ^ wel1 ^ ^ values of X?r in the compartmental lung
model are presented.
D.5. DETERMINATION OF DEPOSITION FRACTIONS
D-8 DRAFT-DO NOT QUOTE OR CITE
-------
The mathematical models for determining the deposition fractions of DEPs in various
regions of the respiratory tract have been developed by Yu and Xu (1986, 1987) and are
adopted in this report. Yu and Xu consider DEPs as a polydisperse aerosol with a specified
mass median aerodynamic diameter (MMAD) and geometrical standard deviation o . Each
diesel particle is represented by a cluster-shaped aggregate within a spherical envelope of dia-
meter de. The envelope diameter dc is related to the aerodynamic diameter of the particle by
the relation
(D-24)
where £ is the bulk density of the particle in g/cm3, Q) = 1 g/cm3; is the packing density,
which is the ratio of the space actually occupied by primary particles in the envelope to the
overall envelope volume; and Cx is the slip factor given by the expression:
Cx = 1 + 2 [1.257 + 0.4 exp -( _ x. )] (D-25)
"x ^
in which X s 8 x 10"6cm3 is the mean free path of air molecules at standard conditions. In
the diesel particle model of Yu and Xu (1986), C, has a value of 1.5 g/cm3 and a <(> value of
0.3 is chosen based upon the best experimental estimates. As a result, Equation D-24 gives
de/da =1.35. In determining the deposition fraction of DEPs, de is used for diffusion and
interception according to the particle model.
D.5.1. DETERMINATION OF (DF)H
Particle deposition in the naso- or oro-pharyngeal region is referred to as head or extra-
thoracic deposition. The amount of particles that enters the lung depends upon the breathing
mode. Normally, more particles are collected via the nasal route than the oral route because
of the nasal hairs and the more complex air passages of the nose. Since the residence time of
diesel particles in the head region during inhalation is very small (about 0.1 second for human
adults at normal breathing), diffusional deposition is insignificant and the major deposition
mechanism is impaction. The following empirical formulas derived by Yu et al. (1981) for
human adults are adopted for deposition prediction of DEPs:
D-9 DRAFT-DO NOT QUOTE OR CITE
-------
For mouth breathing:
(DF)Ht ,„ = 0, for cfe 3000 (D-26)
in = -1-1 17 + °-324 log(0, /or dQ > 3000
/, ex = 0.
and for nose breathing:
(DF>H. in = -°-014 + 0-023 log(^0, for d\Q £ 337 (D-29)
(DF>H, in = -°-959 * 0.397 log(^0, for d\Q > 337 0«0)
= 0.003 - 0.033 log(^0, for d\Q < 215 (D-31)
= -0.851 + 0.399 logfcg), /or daQ > 215 (
where (DF)H is the deposition efficiency in the head, the subscripts in and ex denote inspira-
tion and expiration, respectively, da is the particle aerodynamic diameter in u,m, and Q is the
air flowrate in cm3/sec.
Formulas to calculate deposition of diesel particles in the head region of children are
derived from those for adults using the theory of similarity, which assumes that the air pas-
sage in the head region is geometrically similar for all ages and that the deposition process is
characterized by the Stokes number of the particle. Thus, the set of empirical equations from
D-26 through D-32 are transformed into the following form:
D-10 DRAFT-DO NOT QUOTE OR CITE
-------
For mouth breathing:
(DF)H ,.„ = 0, for d\Q <, 3000 (D-33)
in -1.117 + 0.972
'm (D-34)
0.324 log(0, /or 3000
and for nose breathing:
(DF)H in * - 0.014 + 0.690 log K + 0.023
for d*aQ <. 337
te = -°-959 * i-191 Io8 ^ +
'" - , (D-37)
0.397 log (djg), for tTQ > 337
„ = 0.003 + 0.099 log K
(D-38)
0.033 log(dJ0, for
- 0.851 * 1.197 log K +
0.399 log(0, for d >215
where K is the ratio of the linear dimension of the air passages in the head region of adults to
that of children, which is assumed to be the same as the ratio of adult/child trachea!
diameters.
For rats, the following empirical equations are used for deposition prediction of DEPs in
the nose:
. in . « - 0-046 *
0.009 log(^0, for fa <, 13.33
.- -0.522*
0.514 Iog( 13.33
D-l 1 DRAFT-DO NOT QUOTE OR CITE
-------
D.4.2. Determination of (DF)T and (DF)A
The deposition model adopted for DEPs is the one previously developed for mono-
disperse and (Yu, 1978) and polydisperse spherical aerosols (Diu and Yu, 1983). In the
model, the branching airways are viewed as a chamber model shaped like a trumpet
(Figure D-2). The cross-sectional area of the chamber varies with airway depth, x, measured
from the beginning of the trachea. At the last portion of the trumpet, additional cross-
sectional area is present to account for the alveolar volume per unit length of the airways.
Inhaled diesel panicles that escape capture in the head during inspiration will enter
the trachea and subsequently the bronchial airways (compartment T) and alveolar spaces
(compartment A).
Assuming that the airways expand and contract uniformly during breathing, the
equation for the conservation of particles takes the form:
ftA, + Aj*. + Q ^£ = - Qet\ (D-42)
ax dx
where c is the mean particle concentration at a given x and time t; Aj and A2 are, respec-
tively, the summed cross-sectional area (or volume per unit length) of the airways and alveoli
at rest; r\ is the particle uptake efficiency per unit length of the airway; p is an expansion
factor, given by:
p = 1 + J. (D-43)
and Q is the air flow rate, varying with x and t according to the relation
6 = 1 - -L (D-44)
Qo ^
where Q0 is the air flow rate at x = 0. In Equations D-43 and D-44, Vt is the volume of new
air in the lungs and Vx and V( are, respectively, the accumulated airway volume from x = 0
to x, and total airway volume at rest.
D-12 DRAFT-DO NOT QUOTE OR CITE
-------
Summed Alveolar Croa Sectional Area A,(x)
Trachea
Airway Leagik x
Croa Sectkmil Area A,(x)
Figure D-2.Trumpet model of lung airways.
D-13 DRAFT-DO NOT QUOTE OR CITE
-------
Equation D-42 is solved using the method of characteristics with appropriate initial and
boundary conditions. The amount of particles deposited between location Xj and x2 from time
tj to t2 can then be found from the expression
/2
DF = f f Qcr\dxdt (D-45)
/, or,
For diesel particles, r) is the sum of those due to the individual deposition mechanisms described
above, i.e.,
tl = TI; + ri5 + n + i\D (D-46)
where r)j, t|s, rjp, and D are, respectively, the deposition efficiencies per unit length of the airway
due to impaction, sedimentation, interception, and diffusion. On the basis of the particle model
described above, the expressions for T]J, t)s, TIP, and r|D are obtained in the following form:
tj/ = u-768(g/)fl (D-47)
L
— PE /I - e(2/3> - EvVl - e273 - sin'1 e1/3] d>48)
nl
(D-49)
l[l-0.819exp(-14.63A) -
1 (D-50)
0.0976 exp(-89.22A) -
0.0325 exp(-228A) - 0.0509 exp(
for Reynolds numbers of the flow smaller than 2000, and
- 0.444A1/2)
D-14 DRAFT-DO NOT QUOTE OR CITE
-------
for Reynolds numbers greater than or equal to 2000, where ST=d2au/(18yiR) is the particle Stokes
number, 6 = L/(8R), G = 3nusL/(32uR), T = dJR, and A = DL/(4R2u). In the above definitions
u is the air velocity in the airway; |i is the air viscosity; L and R are, respectively, the length and
radius of the airway; us = CjPJ(18\i) is the particle settling velocity; and D = CekT(^^d^ is
the diffusion coefficient with k denoting the Boltzmann constant and T the absolute temperature.
In the deposition model, it is also assumed that TJJ and t)p = 0 for expiration, while T|D and r\s
have the same expressions for both inspiration and expiration.
During the pause, only diffusion and sedimentation are present. The combined
deposition efficiency in the airway, E, is equal to:
£=!-(!- Eg) (1 - ED) .
(D-52)
where ED and Es are, respectively, the deposition efficiencies due to the individual mechanisms
of diffusion and sedimentation over the pause period. The expression for ED and Es are given
by
D
1 -
1 tt<
exp(- of
.) exp
1/2
E
(D-53)
where TD = Di/R in which T is the pause time and aj
equation:
J0(a) = 0 .
ct2, and a3 are the first three roots of the
(D-54)
in which J0 is the Bessel function of the zeroth order, and:
Es = 1.1094t5 - 0.16044 for 0 < T$ <, 1.
(D-55)
D-15
DRAFT-DO NOT QUOTE OR CITE
-------
and
Es = 1 - 0.0069TJ1 -0.0859T'2 - 0.0582T'3,
for T5 > 1,
where TS
The values of (DF)T and (DF)A over a breathing cycle are calculated by superimposing
DF for inspiration, deposition efficiency E during pause, and DF for expiration in the
tracheobronchial airways and alveolar space. It is assumed that the breathing cycle consists of
a constant flow inspiration, a pause, and a constant flow expiration, each with a respective
duration fraction of 0.435, 0.05, and 0.515 of a breathing period.
D.5.3. Lung Models
Lung architecture affects particle deposition in several ways: the linear dimension of the
airway is related to the distance the particle travels before it contacts the airway surface; the
air flow velocity by which the particles are transported is determined by the cross-section of
the airway for a given volumetric flowrate; and flow characteristics in the airways are
influenced by the airway diameter and branching patterns. Thus, theoretical prediction of
particle deposition depends, to a large extent, on the lung model chosen.
D.5.3.1. Lung Model for Rats - Morphometric data on the lung airways of rats were
reported by Schum and Yeh (1979). Table D-l shows the lung model data for Long Evans
rats with a total lung capacity of 13.784cm3. Application of this model to Fischer rats is
accomplished by assuming that the rat has the same lung structure regardless of its strain and
that the total lung capacity is proportional to the body weight. In addition, it is also assumed
that the lung volume at rest is about 40% of the total lung capacity and that any linear
dimension of the lung is proportional to the cubic root of the lung volume.
D.5.3.2. Lung Model for Human Adults — The lung model of mature human adults
used in the deposition calculation of DEPs is the symmetric lung model developed by Weibel
(1963). In Weibel's model, the airways are assumed to be a dichotomous branching system
with 24 generations. Beginning with the 18th generation, increasing numbers of alveoli are
present on the wall of the airways and the last three generations are completely aleveolated.
Thus, the alveolar region in this model consists of all the airways in the last seven
D-16 DRAFT-DO NOT QUOTE OR CITE
-------
generations. Table D-2 presents the morphometric data of the airways of Weibel's model
adjusted to a total lung volume of 3,000 cm3.
D.5.3.3. Lung Model for Children — The lung model for children in the diesel study
was developed by Yu and Xu (1987) on the basis of available morphometric measurements.
The model assumes a lung structure with dichotomous branching of airways, and it matches
Weibel's model for a subject when evaluated at the age of 25 years, the age at which the lung
is considered to be mature. The number and size of airways as functions of age t (years) are
determined by the following equations:
D.5.3.3.1. Number of airways and alveoli. The number of airways Nj(t) at generation i
for age t is given by
2', for 0 £ / <20 C
A/2,(0 = 221,
jv^l) = Nr(t) -221, for 221 < Nr(t) < 222
N23(t) = 0,
N2l(t) = 221,
2nt for NJf) > 221 + 222, (D-60)
-221 -222
D-17 DRAFT-DO NOT QUOTE OR CITE
-------
Table D-l LUNG MODEL FOR RATS AT TOTAL LUNG CAPACITY
Generation
Number
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16*
17
18
19
21
22
25
24
Number of
Airways
1
2
3
5
8
14
23
38
65
109
184
309
521
877
1,477
2,487
4,974
9,948
19,896
39,792
79,584
318,336
636,672
Length
(cm)
2.680
0.715
0.400
0.176
0.208
0.117
0.114
0.130
0.099
0.091
0.096
0.073
0.075
0.060
0.055
0.035
0.029
0.025
0.022
0.020
0.019
0.017
0.017
Diameter
(cm)
0.340
0.290
0.263
0.203
0.163
0.134
0.123
0.112
0.095
0.087
0.078
0.070
0.058
0.049
0.036
0.020
0.017
0.016
0.015
0.014
0.014
0.014
0.014
Accumulative
Volume1* (cm)
0.243
0.338
0.403
0.431
0.466
0.489
0.520
0.569
0.615
0.674
0.758
0.845
0.948
1.047
1.414
1.185
1.254
1.375
1.595
2.003
2.607
7.554
13.784
'Terminal bronchioles
^Including the attached alveoli volume
(number of alveoli - 3 x 107, alveolar diameter » 0.0086 cm)
D-l 8
DRAFT-DO NOT QUOTE OR CITE
-------
Table D-2 LUNG MODEL BY WEIBEL (1963) ADJUSTED TO 3,000 CM3 LUNG VOLUME
Generation
Number
0
2
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16'
17
18
19
20
21
22
23
Number of
Airways
1
2
4
8
16
32
64
128
256
512
1,024
2,048
4,096
8,192
16,384
32,768
65,536
131,072
262,144
524,283
1,048^76
2,097,152
4,194,304
8,388,608
Length
(cm)
10.260
4.070
1.624
0.650
1.086
0.915
0.769
0.650
0.547
0.462
0393
0333
0282
0.231
0.197
0.171
0.141
0.121
0.100
0.085
0.071
0.060
0.050
0.043
Diameter
(cm)
1.539
1.043
0.710
0.479
0.385
0.299
0.239
0.197
0.159
0.132
0.111
0.093
0.081
0.070
0.063
0.056
0.051
0.046
0.043
0.040
0.038
0.037
0.035
0.035
Accumulative
Volume11 (cm)
19.06
25.63
28.63
29.50
31.69
33.75
35.94
38.38
41.13
44.38
48.25
53.00
59.13
66.25
77.13
90.69
109.25
13931
190.60
288.16
512.94
925.04
1,694.16
3,000.00
Terminal bronchioles
"Including the attached alveolar volume
(number of alveoli = 3 x 103, alveolar diameter = 0.0288 cm)
D-19
DRAFT-DO NOT QUOTE OR CITE
-------
where NT(t) is the total number of airways in the last three airway generations. The empirical
equation for Nr which best fits the available data is
_ , 2.036 x 107(1-0.926*-°150, / < 8
1.468 x 107, t > 8
Thus, Nr(t) increases from approximately 1.5 million at birth to 15 million at 8 years of age
and remains nearly constant thereafter. Equations D-58 to D-60 also imply that in the last
three generations, the airways in the subsequent generation begin to appear only when those in
the preceding generation have completed development.
The number of alveoli as a function of age can be represented by the following equation
according to the observed data:
N^t) = 2.985 x 108(1 -0.919e-°'450 (°-62)
The number of alveoli distributed in the unciliated airways at the airway generation level
is determined by assuming that alveolization of airways takes place sequentially in a proximal
direction. For each generation, alveolization is considered to be complete when the number
of alveoli in that generation reaches the number determined by Weibel's model.
D.5.3.3.2. Airway size. Four sets of data are used to determine airway size during
postnatal growth: (a) total lung volume as a function of age; (b) airway size as given by
Weibel's model; (c) the growth pattern of the bronchial airways; and (d) variation in alveolar
size with age. From these data, it is found that the lung volume, LV(t) at age t, normalized
to Weibel's model at 4,800 cm3 for an adult (25 years old), follows the equation
LV(t) = 0.959 x 105(1 - 0.998e-°°°2/) (cm3).
D-20 DRAFT-DO NOT QUOTE OR CITE
-------
The growth patterns of the bronchial airways are determined by the following equations
- #(25)], (D-64)
Lffi -
(D-65)
where Dj(t) and Lj(t) are, respectively, the airway diameter and length at generation i and age
t, Diw and Liw the corresponding values for Weibel's model, a; and Pj are coefficients given
by
-2.
ttj = 3.26 x lO'^expt-1.183 (i
(D-66)
p, = 1 .OSxlO'6 exp [10.1] (H
\-0.2n
and H(t) is the body height, which varies with age t in the form
H(r) = 1.82 x ltf(l - 0.725c'014r) (cm).
(D-67)
(D-68)
For the growth patterns of the airways in the alveolar region, it is assumed that
— = M for \7
-------
D.6. TRANSPORT RATES
The values of transport rates \%- for rats have been derived from the experimental data
of clearance for diesel soot (Chan et al., 1981; Strom et al., 1987, 1988) and for the particle
associated organics (Sun et al., 1984; Bond et al., 1986; Yu et al. 1991). These values are
used in the present model of lung burden calculation and are listed below:
X - 1.73 (/ = 1,2,3)
(3) - (3) - (3) -
~
- 12 55
~ ^^
= 0.693 (i = 1,2,3)
0.00068 [1 - exp(-0.046w)]
X.- 0.012 exp(-0.
0.00068 exp(-0.046m^62) (i = 1,2,3)
(2) _ .(2) _ .(2) _ .(2) _ 00,29 (D-73)
-HB ~ *-TB ~ "-LB ~ ^AB ~ u-ul/y
-------
0.012 expC-O.ll/w) + 0.00086
* * * * = °-012 exp(-o.
0.00068 exp(-0.046m!62) + 0.0161
*? - *2 + $ * *2 - 0.012 «p(-0.11mi-76) +
0.00068 exp(-0.046m}-62) + 15.7
where X^y is the unit of day"1, and mA S m ^ is the particle burden (in mg) in the alveolar
compartment.
Experimental data on the deposition and clearance of DEPs in humans are not available.
To estimate the lung burden of DEPs for human exposure, it is necessary to extrapolate the
transport rates X^ from rats to humans. For organics, we assume that the transport rates are
the same for rats and humans. This assumption is based upon the observation of Schanker et
al. (1986) that the lung clearance of inhaled lipophilic compounds appears to depend only on
their lipid/water partition coefficients and is independent of species. In contrast, the transport
rates of diesel soot in humans should be different from that of rats, since the alveolar
clearance rate, XA, of insoluble particles at low lung burdens for human adults is
approximately seven times that of rats (Bailey et al., 1982), as previously discussed in Chapter
4.
No data are available on the change of the alveolar clearance rate of insoluble particles in
humans due to excessive lung burdens. It is seen from equation D-79 that X (JJ for rats can be
written in the form
where a, b, c, and d are constants. The right-hand side of equation D-82 consists of two
terms, representing, respectively, macrophage-mediated mechanical clearance and clearance by
dissolution. The first term depends upon the lung burden, whereas the second term does not.
To extrapolate this relationship to humans, we assume that the dissolution clearance term was
independent of species and that the mechanical clearance term for humans varies in the same
D-23 DRAFT-DO NOT QUOTE OR CITE
-------
proportion as in rats under the same unit surface particulate dose. This assumption results in
the following expression forX (!J in humans
tf] = - e\p(-b(mA/Sf) + d (D-83)
where P is a constant derived from the human/rat ratio of the alveolar clearance rate at low
lung burdens, and S is the ratio of the pulmonary surface area between humans and rats.
equation D-83 implies that rats and humans have equivalent amounts of biological response in
the lung to the same specific surface dose of inhaled DEPs.
From the data of Bailey et al. (1982), we obtain a value of X (1J = 0.00169 day'1 for
humans at low lung burdens. This leads to P = 14.4. Also, we find S=148 from the data of
the anatomical lung model of Schum and Yeh (1979) for rats and Weibel's model for human
adults. For humans less than 25 years old, we assume the same value for P, but S is
computed from the data of the lung model for young humans (Yu and Xu 1987). The value
of S for different ages is shown in Table D-3.
The equations for other transport rates that have a lung-burden-dependent component are
extrapolated from rats to humans in a similar manner. The following lists the values of X \
(in day"1) for humans used in the present model calculation:
1.73 (i = 1,2,3)
$B - *$ - 0-00018 (D-85)
- * - 0-0129
(3) - X(3) - X(3) - 12 55
" ~ K ~ K ~ l/>:>:>
D-24 DRAFT-DO NOT QUOTE OR CITE
-------
TABLE D-4. RATIO OF PULMONARY SURFACE AREAS
BETWEEN HUMANS AND RATS AS A FUNCTION OF HUMAN AGE
Age (Year)
0
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
Surface Area
4.99
17.3
27.6
36.7
44.7
51.9
58.5
64.6
70.4
76.0
81.4
86.6
91.6
96.4
101
106
110
115
119
123
128
132
136
140
144
148
D-25 DRAFT-DO NOT QUOTE OR CITE
-------
(£ = 0.693 (i = 1,2,3) (°-88)
1-62
Xjjj- = 0.0694 {0.012 expI-O.ll(myS)1-76] +
0.00068 exp[-0.046(w/1/5)1-76]} (i = 1, 2, 3)
(2) + ,(2) . (2)
- *- h
0.0694{0.012
0.00068 exp[-0.046(m/
-------
To test the accuracy of the model, simulation results are obtained on the retention of
diesel soot in the rat lung and compare with the data of lung burden and lymph node burden
obtained by Strom et al. (1988). A particle size of 0.19 urn MMAD and a standard geometric
deviation, cg of 2.3 (as used in Strom's experiment) are used in the calculation.
The respiratory parameters for rats are based on their weight and calculated using the
following correlations of minute volume, respiratory frequency, and growth curve data.
Minute Volume = 0.9W (cm3/min) (D-95)
Respiratory Frequency = 475W0-3 (1/min) (D-96)
where W is the body weight (in grams) as determined from the equation
W = 5+537T/(100+T), for T£56 days (D-97)
in which T is the age of the rat measured in days
Equation D-95 was obtained from the data of Mauderly (1986) for rats ranging in age
from 3 months to 2 years old; equation D-96 was obtained from the data of Strom et al.
(1988); and equation D-59 was determined from the best fit of the experimental deposition
data. Figures D-3 and D-4 show the calculated lung burden of diesel soot (m flj + mty and
lymph node burden, respectively, for the experiment by Strom et al. (1988) using animals
exposed to DEPs at 6 mg/m3 for 1,3,6 and 12 weeks; exposed in all cases was 7 days/week
and 20 hours daily. The solid lines represent the calculated accumulation of particles during
the continuous exposure phase and the dashed lines indicate calculated post exposure
retention. The agreement between the calculated and the experimental data for both lung and
lymph node burdens during and after the exposure periods was very good.
Comparison of the model calculation and the retention data of particle-associated BaP in
rats obtained by Sun et al. (1984) is shown in Figure D-5. The calculated retention is shown
by the solid line. The experiment of Sun et al. consisted of a 30 minute exposure to diesel
particles coated with [3H\ benzo[a]pyrene (pH] - BaP) at a concentration of 4-6 ug/m3 of air
and followed by a post exposure period of over 25 days. The fast and slow phase of
(I*H] - BaP) clearance half-times were found to be 0.03 day and 18 days, respectively. These
corrospond to \$o = 0.0385 day "l and X %, = 23.1 day"1 in our model, where X ($o is the
value of X $Y at mA -» 0. Figure D-5 shows that the
D-27 DRAFT-DO NOT QUOTE OR CITE
-------
' — — 43— _«. _ ^wk
26 39
Time, week
52
65
Figure D-3.
The Experimental and predicted lung burdens of rats to DEPs at a
solid and dashedconcentration of 0.6 mg/m3 for different exposure
spans.lines are, respectively, the predicted burdens during exposure
and post exposure. Particle characteristics and exposure panern are
explained in the text. The symbols represent the experimental data
from Strom et al. (1988).
D-28
DRAFT-DO NOT QUOTE OR CITE
-------
6r
O)
c
0>
"S
3
CO
CD
•O
O
1 2
Q.
X
X
X
o **
0
••a:
f^-o —• — ~o—
13
26 39
Time, week
52
65
Figure D-4.
Experimental and predicted lymph node burdens of rats exposed to
DEPs at a concentration of 6.0 mg/m3 for different exposure spans.
The solid and dashed lines are, respectively, the predicted burdens
during exposure and post exposure. Particle characteristics and
exposure pattern are explained in the text. The symbols represent the
experimental data from Strom et al. (1988).
D-29
DRAFT-DO NOT QUOTE OR CITE
-------
0.8
c
g
"c
0)
Q)
o:
0.6
)
0.4
0.2
10 15 20
Time, day
25
30
Figure D-5.
Comparison between the calculated lung retention (solid line) and the
experimental data obtained by Sun et al. (1984) for the particle
associated BaP in rats.
D-30 DRAFT-DO NOT QUOTE OR CITE
-------
calculated retention is in excellent agreement with the experimental data obtained by Sun et
al., (1984).
D.7.2. Predicted Burdens in Humans
Selected results of lung burden predictions in humans are shown in Figures D-6 to D-9.
The particle conditions used in the calculation are 0.2 jam MMAD with o*g = 2.3, and themass
fractions of the rapidly and slowly cleared organics are each 10 percent (fj - f2 = 0.1).
Figures D-6 and D-7 show, respectively, the lung burdens per unit concentration of diesel soot
and the associated organics hi human adults for different exposure patterns at two soot
concentrations, 0.1 mg/m3 and 1 mg/m3. The exposure
patterns used in the calculation are (a) 24 hours/day and 7 days week; (b) 12 hours/day and 7
days/week; and (c) 8 hours/day an 5days/week, simulating environmental and occupational
exposure conditions. The results show that the lung burdens of both diesel soot and the
associated organics reached a steady state value during exposure. Due to differences in the
amount of particle intake, the steady state lung burdens per unit concentration were the
highest for exposure pattern (a) and the lowest for exposure pattern (c). Also, increasing soot
concentration from 0.1 mg/m3 to 1 mg/m3 increased the lung burden per unit concentration.
However, the increase was not noticeable for exposure pattern (c). The dependence of lung
burden on the soot concentration is caused by the reduction of the alveolar clearance rate at
high lung burdens discussed above.
Figures D-8 and D-9 show the effect of age on lung burden, where the lung burdens per
unit concentration per unit lung weight are plotted vs. age. The data of lung weight at
different ages are those reported by Snyder (1975). The exposure pattern used in the
calculation is 24 hours/day and 7 days/week for a period of one year at the two soot
concentrations, 0.1 mg/m3 and 1 mg/m3. The results show that, on a unit lung weight basis,
the lung burdens of both soot and the organics are functions of age and the maximum lung
burdens ocurr at approxomately 5 years of age. Again, for any given age, the lung burden
per unit concentration is slightly higher at 1 mg/m3 than at 0.1 mg/m3.
D.8. PARAMETRIC STUDY OF THE MODEL
The deposition and clearance model of DEPs in humans, presented above, consists of a
large number of parameters which characterize: the size and composition of diesel particles,
the structure and dimension of the respiratory tract, the ventilation conditions of the subject,
D-31 DRAFT-DO NOT QUOTE OR CITE
-------
and the clearance half-times of the diesel soot and the particle-associated organics. Any
single or combined changes of these parameters from their normal values in the model would
result in a change in the predicted lung burden. A parametric study has been conducted to
investigate the effects of each individual parameter on calculated lung burden in human
adults. The exposure pattern chosen for this study is 24 hours/day and 7 days/week for a
period of 10 years at a constant soot concentration of 0.1 mg/m3. The following presents two
important results from the parametric study.
4 6
Time, year
10
Figure D-6.
Calculated lung burdens of diesel soot per unit exposure concentration in
human adults exposed continueously to DEPs at two different concentrations
of 0.1 mg/m3 and 1.0 mg/m3. Exposure patterns are (a) 24 hours/day and 7
days/week, (b) 12 hours/day and 7 days/week, and (c) 8 hours/day and 5
days/week.
D-32
DRAFT-DO NOT QUOTE OR CITE
-------
4 6
Time, year
10
Figure D-7.
Calculated lung burdens of the particle-associated organics per unit exposure
concentration in human adults exposed continuously to DEPs at two
different concentrations of 0.1 mg/m3 and 1.0 mg/m3. Exposure patterns are
(a) 24 hours/day and 7 days/week, (b) 12 hours/day and 7 days/week, and
(c) 8 hours/day and 5 days/week.
D-33
DRAFT-DO NOT QUOTE OR CITE
-------
Lung Burden/Lung Weight/Concentration
(mg/g)/(mg/cu. m)
-------
n
u
o
u
r 3
O 0
II
O >.
C O)
I!
O
CD
O
3
0.01
0.008 -
0.006 -
0.004
0.002 -
1 mg/cu. m
10 15
Age, year
20
25
Figure D-9.
Calculated burdens of the particle-associated organics per gram of lung per
unit exposure concentration in humans of different ages exposed
continuously for one year to DEPs of two different concentrations of 0.1
mg/m3 and 1.0 mg/nr for 7 days/week and 24 hours daily.
D-35
DRAFT-DO NOT QUOTE OR CITE
-------
D.8.1. Effect of Ventilation Conditions
The change in lung vurden due to variations in tidal volume and respiratory frequency are
depicted in Figures D-10 and D-ll. Increasing any one of these ventilation parameters
increased the lung burden, but the increase was much smaller with respect to respiratory
frequency than to tidal volume. This small increase in lung burden was a result of the
decrease in deposition efficiency as respiratory frequency increased, despite a higher total
amount of DEPs inhaled.
The mode of breathing has only a minor effect on lung burden because switching from
nose breathing does not produce any appreciable change in the amount of particle intake into
the lung (Yu and Xu 1987). All lung burden results presented in this report are for nose
breathing.
D.8.2. Effect of Tranport Rates
Transport rates have an obvious effect on the retention of DEPs in the lung after
depositon. Because we are mainly concerned with the long-term clearance of diesel soot and
the associated organics, only the effects of two transport rates X ty and X ^ are studied.
Experimental data of X ^ from various diesel studies in rats have shown that X ^ can vary
by a factor of two or higher. We use a multiple of 0.5 to 2 for the uncertainty in X flj and X
^ to examine the effect on lung burden. Figures D-12 and D-13 show respectively, the lung
burden results for diesel soot and the associated organics vs. the multiples of X ^ and X @J
used in the calculation. As expected, increasing the multiple of X ty reduced the lung burden
of diesel soot with practically no change in the organics burden (Figure D-12), while just the
oposite occurred when the multiple of X ^ was increased (Figure D-13).
D-36 DRAFT-DO NOT QUOTE OR CITE
-------
100
o
o
C/)
a>
w
u
'c
re
20 -
0.4
0.5
Tidal Volume, liter
0.6
0.7
Figure D-10.
Calculated lung burdens in human adults vs. tidal volume in liter for
exposure to DEPs at 0.1 mg/m3 for 10 years at 7 days/week and 24 hours
daily. Parameters used in the calculation are: (a) MMAD=0.2 um, o=2.3,
/2""0'1» /3*0-1» O1) respiratory frequency = 14 min"1, and (c) lung volume =
3000 cm3.
D-37
DRAFT-DO NOT QUOTE OR CITE
-------
60
50
40
_-30
o
20
10
Soot
Organics
10
12 14 16
Respiratory Frequency, 1/min
1.4
1.2
0.8
O)
'c
ro
0.4
0.2
18
Figure D-ll.
Calculated lung burdens in human adults vs. respiratory freguency in bpm
for exposure to DEPs at 0.1 mg/m3 for 10 years at 7 days/week and 24
hours daily. Parameters used in the in the calculation are: (a) MMAD=0.2
jim, ag«2.3, /2=0.1, /3=0.1, (b) tidal volume * 500 cm3, and (c) lung
volume = 3200 cm3.
D-38
DRAFT-DO NOT QUOTE OR CITE
-------
120
100
80
—- 60
o
o
40
20
Soot
Organics
2.5
D)
W
'c
O)
6
0.5
0.6 0.8 1 1.2 1.4
Multiple of
1.6
1.8
Figure D-12.
Calculated lung burdens in human adults vs. multiple of X ^for exposure
to DEPs at 0.1 mg/m3 for 10 years at 7 days/week and 24 hours daily.
Parameters used in the calculation are: (a) MMAD=0.2 urn, a.=2.3, 72=0.1,
/3=0.1, (b) tidal volume = 500 cm3, respiratory frequency = 14 min"1, and
(c) lung volume - 3200 cm3.
D-39
DRAFT-DO NOT QUOTE OR CITE
-------
1 1.2 1.4
Multiple of
1.6
1.8
Figure D-13.
Calculated lung burdens in human adults vs. multiple of X ^for exposure to
DEPs at 0.1 mg/m3 for 10 years at 7 days/week and 24 hours daily.
Parameters used in the calculation are: (a) MMAD=0.2 urn a.=2.3, f-fQ.l,
/3=0.1, (b) tidal volume = 500 cm3, respiratory frequency - 14 min , and
(c) lung volume * 3200 cm3.
D-40
DRAFT-DO NOT QUOTE OR CITE
-------
D.9. REFERENCES
Amann, C.A.; Siegla, D.C. (1982) Diesel particles - What are they and why. Aerosol Sci.
Tech. 1:73-101
Bailey, M.R.; Fry, F.A.; James, A.C. (1982) The long-term clearance kinetics of insoluble
particles from the human lung. Ann. Occup. Hyg. 26:273-289.
Bond, J.A.; Sun, J.D.; Medinsky, M.A.; Jones, R.K.; Yeh, H.C. (1986a) Deposition, metabolism
and excretion of l-[14C]nitropyrene and l-[14C]nitropyrene coated on diesel exhaust particles
as influenced by exposure concentration. Toxicol. Appl. Pharmacol. 85:102-117.
Chan, T.L.; Lee, P.S.; Hering, W.E. (1981) Deposition and clearance of inhaled diesel exhaust
particles in the respiratory tract of Fisher rats. J. Appl. Tox. 1:77-82.
Diu, C.K.; Yu, C.P. (1983) Respiratory tract deposition of polydisperse aerosols in humans. Am.
Ind. Hyg. Assoc. J. 44:62-65.
ICRP Publication 30, part 1. 1979. Limits for intakes of radionculides by workers. Ann ICRP 2.
Schanker, L.S.; Mitchell, E.W.; Brown, R.A. (1986) Species comparison of drug absorption from the
lung after aerosol inhalation or intratracheai injection. Drug Metab. Dispos. 14(l):79-88.
Scheutzle, D. (1983) Sampling of vehicle emissions for chemical analysis and biological testing.
Environ. Health Perspect. 47:65-80.
Schum, M.; Yeh, H.C. (1979) Theoretical evaluation of aerosol deposition in anatomical models of
mammalian lung airways. Bull. Math. Biol. 42:1-15.
Snyder WS. 1975. Report of task group on reference man. pp. 151-173. Pergamon Press, Oxford,
London.
Solderholm SC. 1981. Compartmental analysis of diesel particle kinetics in the respiratory sytem of
exposed animals. Oral presentation at EPA Diesel Emissions Symposium, Raleigh, NC, October
5-7. In: Vostal JJ, Schreck RM, Lee PS, Chan TL, Soderholm SC. 1982. Deposition and
clearance of diesel particles from the lung. In: Toxicological Effects of Emissions from Diesel
Engins (Lewtas J, ed.) pp, 143-159. Elsevier, New York, NY.
Strom, K.A.; Chan, T.L.; Johnson, J.T. (1987) Pulmonary retention of inhaled submicron particles in
rats: diesel exhaust exposures and lung retention model. Research Publication GMR-5718,
Warren, MI: General Motors Research Laboratories.
Strom, K.A.; Chan, T.L.; Johnson, J.T. (1988) Inhaled particles VI. Dodgson, J.; McCallum, R.I.;
Bailey, M.R.; Fischer, D.R., eds. London: Pergamon Press, pp. 645-658.
Sun, J.D.; Woff, R.K.; Kanapilly, G.M.; McClellan, R.O. (1984) Lung retention and metabolic fate of
inhaled benzo(a)pyrene associated with diesel exhaust particles. Toxicol. Appl. Pharmacol. 73:48-
59.
D-41 DRAFT-DO NOT QUOTE OR CITE
-------
Weibel, E.R. (1963) Morphometry of the human lung. Berlin:Springer-Verlag.
Yu, C.P. (1978) Exact analysis of aerosol deposition during steady breathing. Powder Tech. 21:55-
62.
Yu, C.P.; Diu, C.K.; Soong, T.T. (1981) Statistical analysis of aerosol deposition in nose and mouth.
Am. Ind. Hyg. Assoc. J. 42:726-733.
Yu, C.P.; Xu, G.B. (1986) Predictive models for deposition of diesel exhaust particiates in human
and rat lungs. Aerosol Sci. Tech. 5:337-347.
Yu, C.P.; Xu, G.B. (1987) Predicted deposition of diesel particles in young humans. J. Aerosol Sci.
18:419-429.
Yu, C.P., Yoon, K.J., Chen, Y.K. (1991) Retention modeling of diesel exhaust particles in rats and
humans. J. Aerosol Medicine. In press.
U.S. Environmental Protection Agency
Region 5, Library (PL-12J)
77 West Jackson Boulevard, 12th Floor
Chicago, IL 60604-3590
D-42 DRAFT-DO NOT QUOTE OR CITE
------- |